.. DO NOT EDIT.
.. THIS FILE WAS AUTOMATICALLY GENERATED BY SPHINX-GALLERY.
.. TO MAKE CHANGES, EDIT THE SOURCE PYTHON FILE:
.. "auto_examples/experiment_custom_lisn.py"
.. LINE NUMBERS ARE GIVEN BELOW.
.. only:: html
.. note::
:class: sphx-glr-download-link-note
:ref:`Go to the end <sphx_glr_download_auto_examples_experiment_custom_lisn.py>`
to download the full example code.
.. rst-class:: sphx-glr-example-title
.. _sphx_glr_auto_examples_experiment_custom_lisn.py:
Implementing and running a custom Experiment on a dataset
==========================================================
This notebook demonstrates implementing a custom experiment to search for hyperparameters and saving the scores of the experiment for a specified set of parameters for `lisn` dataset.
We will inherit a new Experiment class from `cartodata.model_selection.experiment.BaseExperiment` class.
First we will define necessary global variables.
.. GENERATED FROM PYTHON SOURCE LINES 11-28
.. code-block:: Python
from pathlib import Path # noqa
ROOT_DIR = Path.cwd().parent
SOURCE = "authors"
NATURE = "articles"
# The directory where the artifacts of the experiment will be saved
TOP_DIR = ROOT_DIR / "experiment_custom"
# The directory where dataset.yaml files reside
CONF_DIR = ROOT_DIR / "conf"
# The directory where files necessary to load dataset columns reside
INPUT_DIR = ROOT_DIR / "datas"
TOP_DIR
.. rst-class:: sphx-glr-script-out
.. code-block:: none
PosixPath('/builds/2mk6rsew/0/hgozukan/cartolabe-data/experiment_custom')
.. GENERATED FROM PYTHON SOURCE LINES 29-33
Initialize Parameter Iterator
-----------------------------
We will initilialize a parameter iterator to iterate through our parameters. We have two options `GridIterator` and `RandomIterator`.
.. GENERATED FROM PYTHON SOURCE LINES 33-41
.. code-block:: Python
from cartodata.model_selection.iterator import GridIterator, RandomIterator # noqa
help(GridIterator)
""
help(RandomIterator)
.. rst-class:: sphx-glr-script-out
.. code-block:: none
Help on class GridIterator in module cartodata.model_selection.iterator:
class GridIterator(ParameterIteratorBase)
| GridIterator(df=None, csv_filepath=None, params_dict=None, index_col=0)
|
| A parameter iterator class that iterates through a given set of
| parameters in order.
|
| Method resolution order:
| GridIterator
| ParameterIteratorBase
| builtins.object
|
| Methods defined here:
|
| __init__(self, df=None, csv_filepath=None, params_dict=None, index_col=0)
| Initializes the params_frame either by the specified ``df``, from the
| specified ``csv_filepath`` or the ``params_dict``.
|
| Parameters
| ----------
| df: pandas.DataFrame
| the dataframe that contains the list of parameter sets.
| csv_filepath: str, Path
| the path of the .csv file that contains the list of parameter sets.
| params_dict: dict
| dictionary of possible parameters specified as key: list pairs.
| index_col: int
| index column if a csv_filepath is specified.
|
| ----------------------------------------------------------------------
| Methods inherited from ParameterIteratorBase:
|
| has_next(self)
| Checks if there is a set of parameters that is that used yet.
|
| Returns
| -------
| bool
| True if there still is a set of parameters that is not used yet;
| False otherwise.
|
| head(self, n=10)
| Views top `n` entries in the `params_frame.`
|
| Parameters
| ----------
| n: int
| the number of rows to view.
|
| next(self)
| Returns the next set of values as dictionary.
|
| Removes ``selected`` column before returning.
|
| Example set of values:
| {
| "id": 7b5e167fb3242dacae5c,
| "robustseed" : 0,
| "dataset" : "lisn",
| "min_df" : 10,
| "max_df" : 0.5,
| "max_words" : "None",
| "vocab_sample" : "None",
| "filter_min_score" : 3,
| "projectionnD" : "lsa",
| "num_dim" : 50,
| "projection2D": "umap",
| "n_neighbors": 15,
| "min_dist": 0.1,
| "metric": "euclidean",
| "clustering" : "kmeans",
| "base_factor" : 3,
| "weight_name_length" : 0,
| }
|
| set_initial_executed(self)
| Check params_frame for the rows that are already run and sets
| `nb_initial_executed` so that these row are not considered for `nbmax`
| value specified for `stop` function.
|
| stop(self, nbmax=None)
| Checks if stopping condition is satisfied. If the experiment is run
| `nbmax` times or there are no other set of parameters to test, stopping
| condition is satisfied.
|
| Parameters
| ----------
| nbmax: int, default=None
| maximum number of experiments to run.
|
| Returns
| -------
| bool
| True if stopping condition is satisfied; False otherwise.
|
| to_csv(self, filepath)
| Persists `params_frame` to the specified `filepath`.
|
| Parameters
| ----------
| filepath: str, Path
| the path to save the contents of params_frame.
|
| transform(self, x)
|
| ----------------------------------------------------------------------
| Readonly properties inherited from ParameterIteratorBase:
|
| columns
| Returns the column names of `params_frame`.
|
| ----------------------------------------------------------------------
| Data descriptors inherited from ParameterIteratorBase:
|
| __dict__
| dictionary for instance variables (if defined)
|
| __weakref__
| list of weak references to the object (if defined)
Help on class RandomIterator in module cartodata.model_selection.iterator:
class RandomIterator(ParameterIteratorBase)
| RandomIterator(df=None, csv_filepath=None, params_dict=None, index_col=0, seed=None)
|
| A parameter iterator class that iterates through a given set of
| parameters by random selections.
|
| Method resolution order:
| RandomIterator
| ParameterIteratorBase
| builtins.object
|
| Methods defined here:
|
| __init__(self, df=None, csv_filepath=None, params_dict=None, index_col=0, seed=None)
| Initializes the params_frame either by the specified ``df``, from the
| specified ``csv_filepath`` or the ``params_dict``.
|
| Parameters
| ----------
| df: pandas.DataFrame
| the dataframe that contains the list of parameter sets.
| csv_filepath: str, Path
| the path of the .csv file that contains the list of parameter sets.
| params_dict: dict
| dictionary of possible parameters specified as key: list pairs.
| index_col: int
| index column if a csv_filepath is specified.
| seed: int
| seed for random generator
|
| ----------------------------------------------------------------------
| Methods inherited from ParameterIteratorBase:
|
| has_next(self)
| Checks if there is a set of parameters that is that used yet.
|
| Returns
| -------
| bool
| True if there still is a set of parameters that is not used yet;
| False otherwise.
|
| head(self, n=10)
| Views top `n` entries in the `params_frame.`
|
| Parameters
| ----------
| n: int
| the number of rows to view.
|
| next(self)
| Returns the next set of values as dictionary.
|
| Removes ``selected`` column before returning.
|
| Example set of values:
| {
| "id": 7b5e167fb3242dacae5c,
| "robustseed" : 0,
| "dataset" : "lisn",
| "min_df" : 10,
| "max_df" : 0.5,
| "max_words" : "None",
| "vocab_sample" : "None",
| "filter_min_score" : 3,
| "projectionnD" : "lsa",
| "num_dim" : 50,
| "projection2D": "umap",
| "n_neighbors": 15,
| "min_dist": 0.1,
| "metric": "euclidean",
| "clustering" : "kmeans",
| "base_factor" : 3,
| "weight_name_length" : 0,
| }
|
| set_initial_executed(self)
| Check params_frame for the rows that are already run and sets
| `nb_initial_executed` so that these row are not considered for `nbmax`
| value specified for `stop` function.
|
| stop(self, nbmax=None)
| Checks if stopping condition is satisfied. If the experiment is run
| `nbmax` times or there are no other set of parameters to test, stopping
| condition is satisfied.
|
| Parameters
| ----------
| nbmax: int, default=None
| maximum number of experiments to run.
|
| Returns
| -------
| bool
| True if stopping condition is satisfied; False otherwise.
|
| to_csv(self, filepath)
| Persists `params_frame` to the specified `filepath`.
|
| Parameters
| ----------
| filepath: str, Path
| the path to save the contents of params_frame.
|
| transform(self, x)
|
| ----------------------------------------------------------------------
| Readonly properties inherited from ParameterIteratorBase:
|
| columns
| Returns the column names of `params_frame`.
|
| ----------------------------------------------------------------------
| Data descriptors inherited from ParameterIteratorBase:
|
| __dict__
| dictionary for instance variables (if defined)
|
| __weakref__
| list of weak references to the object (if defined)
.. GENERATED FROM PYTHON SOURCE LINES 42-43
We define the set of parameters that we want to test.
.. GENERATED FROM PYTHON SOURCE LINES 43-60
.. code-block:: Python
from cartodata.phases import PhaseProjectionND, PhaseProjection2D, PhaseClustering
params = {
"robustseed" : [0],
"authors__filter_min_score": [4],
"filter_min_score": [6],
PhaseProjectionND.NAME : [
{ "key": ["lsa"], "num_dims": [50, 100], "extra_param": [True, False]},
{ "key": ["bert"], "family": ["all-MiniLM-L6-v2"]}
],
PhaseProjection2D.NAME : [
{ "key": ["umap"], "n_neighbors" : [10, 20, 50], "min_dist" : [0.1, 0.25, 0.5], "metric" : ["euclidean"] }
]
}
.. GENERATED FROM PYTHON SOURCE LINES 61-64
The `params` dictionary contains the parameters that will be used for generating matrices, projections, etc. For this experiment, we are going to use `cartodata.pipeline.datasets` module to create a dataset instance, `cartodata.pipeline.projectionnd` and `cartodata.pipeline.projection2d` modules for projections and `cartodata.pipeline.clustering` to generate clusters. The listed parameters are necessary for the constructors of the classes that we are going to use. Projection and clustering classes in these modules extend `cartodata.pipeline.base.BaseEntity` class that provides us `params` property that returns the `key` and certain parameter values of the entity. This property can be used to generate hierarchical directory structure corresponding to the entity/estimator used.
In the params dictionary above, all defined fields exist for the classes in the modules specified above, except `extra_param`. We will also demonstrate how we can use this parameter to affect the directory structure. We will use this parameter for n-dimensional projection.
.. GENERATED FROM PYTHON SOURCE LINES 64-70
.. code-block:: Python
param_iterator = GridIterator(params_dict=params)
""
param_iterator.params_frame.shape
.. rst-class:: sphx-glr-script-out
.. code-block:: none
(45, 7)
.. GENERATED FROM PYTHON SOURCE LINES 71-75
Implement custom Scoring
--------------------------------------
All possible scoring classes are in the `cartodata.model_selection.scoring` module. We will run scoring for each, so we will import them all.
.. GENERATED FROM PYTHON SOURCE LINES 75-81
.. code-block:: Python
from cartodata.model_selection.scoring import (
NeighborsND, Neighbors2D, Comparative, TrustworthinessSklearn,
TrustworthinessUmap, Clustering, FinalScore
)
.. GENERATED FROM PYTHON SOURCE LINES 82-90
Besides we can define a new Custom score. When defining a new score, there are some points that we have to pay attention:
- The new scoring should inherit from `cartodata.model_selection.scoring.ScoringBase` class.
- It should define `KEY` as class variable.
- It should implement an `evaluate` function as `classmethod`
- `evaluate` function should return a `cartodata.model_selection.utils.Result` instance
- If the parameters of the scoring is to be listed in the final results, scoring parameters should be appended to the result
.. GENERATED FROM PYTHON SOURCE LINES 90-114
.. code-block:: Python
from cartodata.model_selection.scoring import ScoringBase
from cartodata.model_selection.utils import Result
class CustomScore(ScoringBase):
KEY = "custom"
DEFAULTS = {
"factor": 20
}
@classmethod
def evaluate(cls, key_nD, factor=20):
result = Result()
# we can assume that key_nD is necessary to calculate this score.
# factor is a scoring variable that can change the value of the scoring,
# it will be specified as kwargs
result.add_score(f"custom_score_{key_nD}", factor)
print(f"{cls.KEY} scores:\n{result.print()}")
return result
.. GENERATED FROM PYTHON SOURCE LINES 115-124
Implement custom Experiment class
--------------------------------------
`cartodata.model_selection.experiment.BaseExperiment` class implements necessary functions to run an experiment. However it does not implement the `run_step` function to execute steps of each iteration; generation of matrices for each phase and the scoring of the results.
`cartodata.pipeline.experiment.PipelineExperiment` is one example class that inherits `BaseExperiment` and implements `run_steps` function using Pipeline API.
Now we will implement a custom experiment. First let's have a look at the documentation for `BaseExperiment`.
.. GENERATED FROM PYTHON SOURCE LINES 124-129
.. code-block:: Python
from cartodata.model_selection.experiment import BaseExperiment # noqa
help(BaseExperiment)
.. rst-class:: sphx-glr-script-out
.. code-block:: none
Help on class BaseExperiment in module cartodata.model_selection.experiment:
class BaseExperiment(builtins.object)
| BaseExperiment(dataset_name, dataset_version, top_dir, conf_dir, input_dir, nature, source, selector, score_list=None, **score_params)
|
| A base class for an experiment.
|
| Methods defined here:
|
| __init__(self, dataset_name, dataset_version, top_dir, conf_dir, input_dir, nature, source, selector, score_list=None, **score_params)
| Parameters
| ----------
| nature: str
| source: str
| top_dir: str or Path
| the top directory inside which experiment artifacts will be
| generated
| selector: cartodata.model_selection.selector.ParameterSelector
| the selector that provides the next set of parameters for the
| experiment
| score_list: list of cartodata.model_selection.evaluators.Evaluators,
| default=None
| score_params: kwargs
| The parameters that will be used by evaluators should be specified
| as kwargs in the format ``evaluator_key__parameter_name=value``
| For example ``name_list`` parameter for ``final_score`` should be
| specified as:
|
| ``final_score__name_list=["neighbors_nd_articles_authors",
| "neighbors_2d_articles_authors",
| "clu_score"]
|
| add_2D_scores(self, all_natures, dir_mat, key_nD, key_2D, dir_nD, dir_2D, plots=None, run_params={})
|
| add_clustering_scores(self, all_natures, cluster_natures, key_nD, key_2D, dir_mat, dir_nD, dir_2D, dir_clus, words_index, plots=None, run_params={})
|
| add_nD_scores(self, key_nD=None, dir_mat=None, dir_nD=None, run_params={})
|
| add_plots_to_scoring(self, plots, natures)
|
| add_post_scores(self, dir_clus, run_params={})
|
| finish_iteration(self, next_params, dir_clus)
|
| persist_result(self, result, phase, phase_result, dump_dir)
|
| run(self, nbmax)
| Runs the experiment for the specified `nbmax` number of times.
|
| :param nbmax: maximum number of experiments to run.
|
| run_steps(next_params)
|
| save_plots(self, natures, matrices, dir_dump, title_parts, annotations=None, annotation_mat=None, file_ext='.png')
|
| ----------------------------------------------------------------------
| Data descriptors defined here:
|
| __dict__
| dictionary for instance variables (if defined)
|
| __weakref__
| list of weak references to the object (if defined)
.. GENERATED FROM PYTHON SOURCE LINES 130-131
We extend the `BaseExperiment` class and implement `run_steps` function.
.. GENERATED FROM PYTHON SOURCE LINES 131-351
.. code-block:: Python
from cartodata.operations import (
dump_matrices, dump_scores, load_matrices_from_dumps, load_scores,
dump_labels, load_labels
)
from cartodata.phases import (
PhaseProjectionND, PhaseProjection2D, PhaseClustering, PhasePost
)
class Experiment(BaseExperiment):
# assume we want to add custom score to nD phase
# we can override add_nD_scores to add the new scoring to this phase
def add_nD_scores(self, key_nD=None, dir_mat=None, dir_nD=None,
run_params={}):
# call super class definition so that the previously defined scorings
# for this phase run
phase_result = super().add_nD_scores(key_nD, dir_mat, dir_nD, run_params)
phase_result.print()
# run our custom scoring
if self._score_exists_in_list(CustomScore):
custom_score_params = CustomScore.get_kwargs(self.score_params)
result = CustomScore.evaluate(
key_nD, **custom_score_params
)
result.append_run_params(run_params)
result.append_score_params(custom_score_params)
self.persist_result(result, PhaseProjectionND, phase_result,
dir_nD)
return phase_result
def _entity_matrices_exist(self, natures, matrix_type, working_dir=None):
if natures is None or len(natures) == 0:
print("No entity natures specified!")
for nature in natures:
entity_files = working_dir.rglob(
f"{nature}_{matrix_type}.*"
)
if len(list(entity_files)) == 0:
return False
print(
f"Matrices of type {matrix_type} for {', '.join(natures)}"
" already exist. Skipping creation."
)
return True
def run_steps(self, next_params):
print(f"\nRunning experiment for parameters:\n{next_params}")
#-------------------------------------
# load dataset
dataset = self._load_dataset(next_params)
dataset.input_dir = self.input_dir
dataset.update_top_dir(self.top_dir)
# set top working dir for next_params
robustseed = next_params["robustseed"]
dir_mat = dataset.working_dir / str(robustseed) / dataset.params
dir_mat.mkdir(parents=True, exist_ok=True)
#-------------------------------------
# create and save entity matrices
if self._entity_matrices_exist(dataset.natures, "mat", dir_mat):
matrices = load_matrices_from_dumps(dataset.natures, "mat", dir_mat)
scores = load_scores(dataset.natures, dir_mat)
else:
matrices, scores = dataset.create_matrices_and_scores()
dump_scores(dataset.natures, scores, dir_mat)
dump_matrices(dataset.natures, matrices, "mat", dir_mat)
#-------------------------------------
# do n-dimensional projection
projection_nD = PhaseProjectionND.get_executor(next_params)
key_nD = projection_nD.key
# First get the value of extra_param
dict_projection_nD = next_params.get(PhaseProjectionND.NAME)
extra_param = dict_projection_nD.get("extra_param", None)
if extra_param is not None:
# if it is required to use this parameter for nD projection,
# the relevant class should also be modified to set self.extra_param
# set this value to params of projection_nD instance
projection_nD._add_to_params(["extra_param"])
projection_nD.extra_param = extra_param
dir_nD = dir_mat / projection_nD.params
# The value of the extra_param will appear in the name of the
# directory generated for nD projection.
dir_nD.mkdir(parents=True, exist_ok=True)
if self._entity_matrices_exist(dataset.natures,
key_nD, dir_nD):
matrices_nD = load_matrices_from_dumps(dataset.natures, key_nD, dir_nD)
else:
matrices_nD = projection_nD.execute(matrices, dataset, dir_nD)
print(f'{key_nD} matrices generated.')
dump_matrices(dataset.natures, matrices_nD, key_nD, dir_nD)
# add n-dimensional projection scores
phase_result = self.add_nD_scores(
key_nD, dir_mat, dir_nD, run_params=projection_nD.params_values
)
self.result.append(phase_result)
self.result.persist_scores(dir_nD)
print(
f"{PhaseProjectionND.long_name()} scores:\n{phase_result.print()}"
)
#-------------------------------------
# do 2-dimensional projection
projection_2D = PhaseProjection2D.get_executor(next_params)
key_2D = projection_2D.key
dir_2D = dir_nD / projection_2D.params
dir_2D.mkdir(parents=True, exist_ok=True)
if self._entity_matrices_exist(dataset.natures,
key_2D, dir_2D):
matrices_2D = load_matrices_from_dumps(dataset.natures, key_2D, dir_2D)
else:
matrices_2D = projection_2D.execute(matrices_nD, dir_2D)
print(f'{key_2D} matrices generated.')
dump_matrices(dataset.natures, matrices_2D, key_2D, dir_2D)
# save 2D plots
title_parts = [dataset.name,
dataset.version,
projection_nD.params,
projection_2D.params]
fig = self.save_plots(dataset.natures, matrices_2D, dir_2D, title_parts)
# add 2-dimensional projection scores
phase_result = self.add_2D_scores(
dataset.natures, dir_mat, key_nD, key_2D, dir_nD,
dir_2D, plots=[fig], run_params=projection_2D.params_values
)
self.result.append(phase_result)
self.result.persist_scores(dir_2D)
print(
f"{PhaseProjection2D.long_name()} scores:\n{phase_result.print()}"
)
#-------------------------------------
# do clustering
clustering = PhaseClustering.get_executor(next_params)
cluster_natures = clustering.natures
key_clus = clustering.key
dir_clus = dir_2D / clustering.params
dir_clus.mkdir(parents=True, exist_ok=True)
if (self._entity_matrices_exist(cluster_natures, key_2D, dir_clus) and
self._entity_matrices_exist(cluster_natures, key_nD, dir_clus)):
clus_scores = load_scores(cluster_natures, dir_clus)
clus_nD = load_matrices_from_dumps(cluster_natures, key_nD, dir_clus)
clus_2D = load_matrices_from_dumps(cluster_natures, key_2D, dir_clus)
clus_eval_pos = load_scores(cluster_natures, dir_clus,
suffix="eval_pos")
clus_eval_neg = load_scores(cluster_natures, dir_clus,
suffix="eval_neg")
labels = load_labels(cluster_natures, dir_clus)
else:
(clus_nD, clus_2D, clus_scores,
cluster_labels, cluster_eval_pos,
cluster_eval_neg) = clustering.create_clusters(
matrices, matrices_2D, matrices_nD, scores, dataset.corpus_index
)
dump_scores(cluster_natures, clus_scores, dir_clus)
dump_matrices(cluster_natures, clus_nD, key_nD, dir_clus)
dump_matrices(cluster_natures, clus_2D, key_2D, dir_clus)
dump_scores(cluster_natures, cluster_eval_pos, dir_clus, suffix="eval_pos")
dump_scores(cluster_natures, cluster_eval_neg, dir_clus, suffix="eval_neg")
dump_labels(cluster_natures, cluster_labels, dir_clus)
# save clustering plots
figs = []
for i, nature in enumerate(cluster_natures):
clus_scores_i = clus_scores[i]
clus_mat_i = clus_2D[i]
title_parts = [dataset.name,
dataset.version,
projection_nD.params,
projection_2D.params,
clustering.key,
nature]
fig = self.save_plots(
dataset.natures, matrices_2D, dir_clus, title_parts,
annotations=clus_scores_i.index, annotation_mat=clus_mat_i
)
figs.append(fig)
# add clustering scores
phase_result = self.add_clustering_scores(
dataset.natures, cluster_natures, key_nD, key_2D,
dir_mat, dir_nD, dir_2D, dir_clus, dataset.corpus_index,
plots=figs, run_params=clustering.params_values
)
self.result.append(phase_result)
self.result.persist_scores(dir_clus)
print(
f"{PhaseClustering.long_name()} scores:\n{phase_result.print()}"
)
#-------------------------------------
phase_result = self.add_post_scores(dir_clus)
self.result.append(phase_result)
self.result.persist_scores(dir_clus)
print(
f"{PhasePost.long_name()} scores:\n{phase_result.print()}"
)
self.finish_iteration(next_params, dir_clus)
print("Finished running step!")
.. GENERATED FROM PYTHON SOURCE LINES 352-362
Now we can initialize `Experiment` instance to run our scoring.
We need to specify which scores to calculate to the experiment using the parameter `score_list`. If we do not specify, the experiment will evaluate scores for all available scoring classes.
We will do it explicitly and specify all classes defined in `cartodata.model_selection.scoring` module, plus `CustomScore` class that we have defined above.
It is possible to specify parameters for score classes as keyword arguments.
For example, if `cartodata.model_selection.scoring.FinalScore` is specified in the `score_list`, the experiment calculates an aggregated score taking average of all scores at the end of each run. If instead of all scores, we want a subset of scores to be included in the average, we can specify it using `final_score__name_list`. For each scoring class, we should name the parameter in the format `scoring_KEY__scoring_parameter`.
.. GENERATED FROM PYTHON SOURCE LINES 362-386
.. code-block:: Python
from cartodata.phases import PhaseProjectionND, PhaseProjection2D, PhaseClustering
experiment = Experiment(
"lisn", "2022.11.15.1", TOP_DIR, CONF_DIR, INPUT_DIR,
NATURE, SOURCE, param_iterator,
score_list=[CustomScore,
NeighborsND,
Neighbors2D,
Comparative,
# TrustworthinessSklearn,
# TrustworthinessUmap,
Clustering,
FinalScore],
final_score__name_list=[
PhaseProjectionND.prefix("neighbors_articles_authors"),
PhaseProjection2D.prefix("neighbors_articles_authors"),
PhaseClustering.prefix("clu_score")],
neighbors__recompute=True,
neighbors__min_score =30,
# trustworthiness_sklearn__n_neighbors=10,
custom__factor=10
)
.. GENERATED FROM PYTHON SOURCE LINES 387-402
We have initialized the experiment and specified
- ```
name_list=[
PhaseProjectionND.prefix("neighbors_articles_authors"),
PhaseProjection2D.prefix("neighbors_articles_authors"),
PhaseClustering.prefix("clu_score")
]
```
for `FinalScore` scoring
- `recompute=True`, `min_score=30` for `NeighborsND` and `Neighbors2D` scoring
- `n_neighbors=10` for `TrustworthinessSklearn`
- `factor=10` for `CustomScore`
Now we will run the experiment for 3 different set of parameters.
.. GENERATED FROM PYTHON SOURCE LINES 402-405
.. code-block:: Python
results = experiment.run(3)
.. rst-class:: sphx-glr-script-out
.. code-block:: none
Running experiment for parameters:
{'id': '68d289ffd85ae917e710', 'robustseed': 0, 'authors__filter_min_score': 4, 'filter_min_score': 6, 'projection_nD': {'key': 'lsa', 'num_dims': 50, 'extra_param': True}, 'projection_2D': {'key': 'umap', 'n_neighbors': 10, 'min_dist': 0.1, 'metric': 'euclidean'}}
lsa matrices generated.
================================
Run params : {'2_nD__rkey': 'lsa', '2_nD__rnum_dims': 50, '2_nD__rnormalize': True, '2_nD__rextra_param': True}
--------------------------------
Scoring params : {'2_nD__smin_score': 30, '2_nD__srecompute': True, '2_nD__ssample_size': None, '2_nD__sn_neighbors': 10, '2_nD__srandom_state': 42}
================================
----------- Scores -----------
2_nD__neighbors_articles_authors : 0.4779
--------- Desc Scores ---------
2_nD__neighbors_articles_authors_det
0 Céline Gicquel : 0.82
1 Loïc Paulevé : 0.73
2 Nikolaus Hansen : 0.69
3 Sébastien Tixeuil : 0.68
4 Philippe Caillou : 0.67
5 Cyril Furtlehner : 0.67
6 Raymond Ros : 0.66
7 Isabelle Guyon : 0.62
8 Dimo Brockhoff : 0.61
9 Paola Tubaro : 0.60
10 Anne Auger : 0.59
11 Yann Ponty : 0.59
12 Sarah Cohen-Boulakia : 0.58
13 Olivier Teytaud : 0.57
14 Nicolas Bredeche : 0.54
15 Marc Baboulin : 0.54
16 Tobias Isenberg : 0.54
17 Albert Cohen : 0.53
18 Guillaume Charpiat : 0.52
19 Chantal Reynaud : 0.51
20 Fatiha Saïs : 0.51
21 Ioana Manolescu : 0.51
22 Sylvie Boldo : 0.49
23 Lonni Besançon : 0.49
24 Nathalie Pernelle : 0.48
25 Franck Cappello : 0.48
26 Petra Isenberg : 0.48
27 Guillaume Melquiond : 0.47
28 Claude Marché : 0.47
29 Marc Schoenauer : 0.44
30 Jean-Daniel Fekete : 0.44
31 François Goasdoué : 0.43
32 Fatiha Zaidi : 0.39
33 Olivier Chapuis : 0.38
34 Michèle Sebag : 0.38
35 Nathann Cohen : 0.38
36 Steven Martin : 0.36
37 Anastasia Bezerianos : 0.32
38 Pierre Dragicevic : 0.32
39 Wendy Mackay : 0.30
40 Michel Beaudouin-Lafon : 0.30
41 Alain Denise : 0.30
42 Balázs Kégl : 0.29
43 Caroline Appert : 0.27
44 Evelyne Lutton : 0.27
45 Johanne Cohen : 0.27
46 Emmanuel Pietriga : 0.26
47 Wendy E. Mackay : 0.20
dtype: object
VALUE : 0.4779
--------- Raw Scores ---------
2_nD__neighbors_articles_authors
Sébastien Tixeuil 0.680851
Michèle Sebag 0.383942
Johanne Cohen 0.272093
Albert Cohen 0.529310
Wendy E. Mackay 0.200000
Philippe Caillou 0.674419
Alain Denise 0.297222
Jean-Daniel Fekete 0.437107
Emmanuel Pietriga 0.260317
Yann Ponty 0.586364
Marc Schoenauer 0.443165
Franck Cappello 0.480488
Caroline Appert 0.273913
Michel Beaudouin-Lafon 0.297531
Wendy Mackay 0.302128
Anne Auger 0.591139
Evelyne Lutton 0.272973
Pierre Dragicevic 0.317284
Ioana Manolescu 0.507317
Nikolaus Hansen 0.686420
Nicolas Bredeche 0.544118
Olivier Teytaud 0.567925
François Goasdoué 0.430189
Nathalie Pernelle 0.482353
Fatiha Saïs 0.507317
Sarah Cohen-Boulakia 0.581818
Claude Marché 0.465957
Chantal Reynaud 0.513333
Olivier Chapuis 0.384615
Steven Martin 0.356410
Fatiha Zaidi 0.390625
Balázs Kégl 0.286842
Paola Tubaro 0.602564
Raymond Ros 0.655882
Cyril Furtlehner 0.666667
Anastasia Bezerianos 0.319403
Sylvie Boldo 0.488571
Guillaume Melquiond 0.472727
Marc Baboulin 0.540000
Dimo Brockhoff 0.610256
Nathann Cohen 0.380488
Petra Isenberg 0.479439
Tobias Isenberg 0.536752
Loïc Paulevé 0.726190
Céline Gicquel 0.818421
Isabelle Guyon 0.623596
Guillaume Charpiat 0.522581
Lonni Besançon 0.487879
dtype: float64
================================
Run params : {}
--------------------------------
Scoring params : {}
================================
----------- Scores -----------
custom_score_lsa : 10.0000
--------- Desc Scores ---------
--------- Raw Scores ---------
custom scores:
None
================================
Run params : {'2_nD__rkey': 'lsa', '2_nD__rnum_dims': 50, '2_nD__rnormalize': True, '2_nD__rextra_param': True}
--------------------------------
Scoring params : {'2_nD__smin_score': 30, '2_nD__srecompute': True, '2_nD__ssample_size': None, '2_nD__sn_neighbors': 10, '2_nD__srandom_state': 42, '2_nD__sfactor': 10}
================================
----------- Scores -----------
2_nD__neighbors_articles_authors : 0.4779
2_nD__custom_score_lsa : 10.0000
--------- Desc Scores ---------
2_nD__neighbors_articles_authors_det
0 Céline Gicquel : 0.82
1 Loïc Paulevé : 0.73
2 Nikolaus Hansen : 0.69
3 Sébastien Tixeuil : 0.68
4 Philippe Caillou : 0.67
5 Cyril Furtlehner : 0.67
6 Raymond Ros : 0.66
7 Isabelle Guyon : 0.62
8 Dimo Brockhoff : 0.61
9 Paola Tubaro : 0.60
10 Anne Auger : 0.59
11 Yann Ponty : 0.59
12 Sarah Cohen-Boulakia : 0.58
13 Olivier Teytaud : 0.57
14 Nicolas Bredeche : 0.54
15 Marc Baboulin : 0.54
16 Tobias Isenberg : 0.54
17 Albert Cohen : 0.53
18 Guillaume Charpiat : 0.52
19 Chantal Reynaud : 0.51
20 Fatiha Saïs : 0.51
21 Ioana Manolescu : 0.51
22 Sylvie Boldo : 0.49
23 Lonni Besançon : 0.49
24 Nathalie Pernelle : 0.48
25 Franck Cappello : 0.48
26 Petra Isenberg : 0.48
27 Guillaume Melquiond : 0.47
28 Claude Marché : 0.47
29 Marc Schoenauer : 0.44
30 Jean-Daniel Fekete : 0.44
31 François Goasdoué : 0.43
32 Fatiha Zaidi : 0.39
33 Olivier Chapuis : 0.38
34 Michèle Sebag : 0.38
35 Nathann Cohen : 0.38
36 Steven Martin : 0.36
37 Anastasia Bezerianos : 0.32
38 Pierre Dragicevic : 0.32
39 Wendy Mackay : 0.30
40 Michel Beaudouin-Lafon : 0.30
41 Alain Denise : 0.30
42 Balázs Kégl : 0.29
43 Caroline Appert : 0.27
44 Evelyne Lutton : 0.27
45 Johanne Cohen : 0.27
46 Emmanuel Pietriga : 0.26
47 Wendy E. Mackay : 0.20
dtype: object
VALUE : 0.4779
--------- Raw Scores ---------
2_nD__neighbors_articles_authors
Sébastien Tixeuil 0.680851
Michèle Sebag 0.383942
Johanne Cohen 0.272093
Albert Cohen 0.529310
Wendy E. Mackay 0.200000
Philippe Caillou 0.674419
Alain Denise 0.297222
Jean-Daniel Fekete 0.437107
Emmanuel Pietriga 0.260317
Yann Ponty 0.586364
Marc Schoenauer 0.443165
Franck Cappello 0.480488
Caroline Appert 0.273913
Michel Beaudouin-Lafon 0.297531
Wendy Mackay 0.302128
Anne Auger 0.591139
Evelyne Lutton 0.272973
Pierre Dragicevic 0.317284
Ioana Manolescu 0.507317
Nikolaus Hansen 0.686420
Nicolas Bredeche 0.544118
Olivier Teytaud 0.567925
François Goasdoué 0.430189
Nathalie Pernelle 0.482353
Fatiha Saïs 0.507317
Sarah Cohen-Boulakia 0.581818
Claude Marché 0.465957
Chantal Reynaud 0.513333
Olivier Chapuis 0.384615
Steven Martin 0.356410
Fatiha Zaidi 0.390625
Balázs Kégl 0.286842
Paola Tubaro 0.602564
Raymond Ros 0.655882
Cyril Furtlehner 0.666667
Anastasia Bezerianos 0.319403
Sylvie Boldo 0.488571
Guillaume Melquiond 0.472727
Marc Baboulin 0.540000
Dimo Brockhoff 0.610256
Nathann Cohen 0.380488
Petra Isenberg 0.479439
Tobias Isenberg 0.536752
Loïc Paulevé 0.726190
Céline Gicquel 0.818421
Isabelle Guyon 0.623596
Guillaume Charpiat 0.522581
Lonni Besançon 0.487879
dtype: float64
2_projection_nD scores:
None
umap matrices generated.
================================
Run params : {'3_2D__rkey': 'umap', '3_2D__rmetric': 'euclidean', '3_2D__rn_neighbors': 10, '3_2D__rmin_dist': 0.1, '3_2D__rinit': 'random', '3_2D__rlearning_rate': 1.0, '3_2D__rn_epochs': None, '3_2D__rrandom_state': None}
--------------------------------
Scoring params : {'3_2D__smin_score': 30, '3_2D__srecompute': True, '3_2D__ssample_size': None, '3_2D__sn_neighbors': 10, '3_2D__srandom_state': 42}
================================
----------- Scores -----------
3_2D__neighbors_articles_authors : 0.4324
3_2D__2_nD_3_2D_neighbors_articles_authors_dim_mean : -0.0455
3_2D__2_nD_3_2D_neighbors_articles_authors_dim_median : -0.0446
--------- Desc Scores ---------
3_2D__neighbors_articles_authors_det
0 Loïc Paulevé : 0.77
1 Céline Gicquel : 0.66
2 Nikolaus Hansen : 0.61
3 Dimo Brockhoff : 0.61
4 Isabelle Guyon : 0.58
5 Cyril Furtlehner : 0.57
6 Tobias Isenberg : 0.57
7 Sébastien Tixeuil : 0.56
8 Yann Ponty : 0.56
9 Philippe Caillou : 0.56
10 Raymond Ros : 0.54
11 Anne Auger : 0.53
12 Albert Cohen : 0.51
13 Guillaume Melquiond : 0.51
14 Paola Tubaro : 0.50
15 Olivier Teytaud : 0.49
16 Nicolas Bredeche : 0.49
17 Sarah Cohen-Boulakia : 0.49
18 Marc Baboulin : 0.48
19 Fatiha Saïs : 0.47
20 Chantal Reynaud : 0.47
21 Sylvie Boldo : 0.47
22 Franck Cappello : 0.46
23 Nathann Cohen : 0.45
24 Ioana Manolescu : 0.44
25 Lonni Besançon : 0.43
26 Marc Schoenauer : 0.42
27 Jean-Daniel Fekete : 0.41
28 Petra Isenberg : 0.41
29 Nathalie Pernelle : 0.41
30 Guillaume Charpiat : 0.40
31 Michèle Sebag : 0.39
32 Fatiha Zaidi : 0.38
33 Claude Marché : 0.37
34 Steven Martin : 0.35
35 Olivier Chapuis : 0.32
36 Anastasia Bezerianos : 0.31
37 François Goasdoué : 0.31
38 Pierre Dragicevic : 0.31
39 Alain Denise : 0.29
40 Michel Beaudouin-Lafon : 0.28
41 Johanne Cohen : 0.26
42 Caroline Appert : 0.24
43 Wendy Mackay : 0.23
44 Emmanuel Pietriga : 0.23
45 Balázs Kégl : 0.23
46 Evelyne Lutton : 0.23
47 Wendy E. Mackay : 0.21
dtype: object
VALUE : 0.4324
3_2D__2_nD_3_2D_neighbors_articles_authors_dim_det
0 Céline Gicquel : -0.16
1 Guillaume Charpiat : -0.13
2 Raymond Ros : -0.12
3 François Goasdoué : -0.12
4 Sébastien Tixeuil : -0.12
5 Philippe Caillou : -0.12
6 Paola Tubaro : -0.11
7 Cyril Furtlehner : -0.09
8 Sarah Cohen-Boulakia : -0.09
9 Claude Marché : -0.09
10 Olivier Teytaud : -0.07
11 Nathalie Pernelle : -0.07
12 Nikolaus Hansen : -0.07
13 Petra Isenberg : -0.07
14 Wendy Mackay : -0.07
15 Olivier Chapuis : -0.07
16 Ioana Manolescu : -0.06
17 Anne Auger : -0.06
18 Lonni Besançon : -0.06
19 Marc Baboulin : -0.06
20 Balázs Kégl : -0.06
21 Nicolas Bredeche : -0.05
22 Isabelle Guyon : -0.05
23 Evelyne Lutton : -0.05
24 Chantal Reynaud : -0.04
25 Caroline Appert : -0.04
26 Fatiha Saïs : -0.03
27 Yann Ponty : -0.03
28 Emmanuel Pietriga : -0.03
29 Jean-Daniel Fekete : -0.02
30 Marc Schoenauer : -0.02
31 Franck Cappello : -0.02
32 Sylvie Boldo : -0.02
33 Albert Cohen : -0.02
34 Fatiha Zaidi : -0.02
35 Michel Beaudouin-Lafon : -0.01
36 Johanne Cohen : -0.01
37 Steven Martin : -0.01
38 Pierre Dragicevic : -0.01
39 Alain Denise : -0.01
40 Anastasia Bezerianos : -0.01
41 Dimo Brockhoff : -0.01
42 Michèle Sebag : 0.00
43 Wendy E. Mackay : 0.01
44 Guillaume Melquiond : 0.03
45 Tobias Isenberg : 0.03
46 Loïc Paulevé : 0.04
47 Nathann Cohen : 0.07
dtype: object
VALUE : -0.0455
--------- Raw Scores ---------
3_2D__neighbors_articles_authors
Sébastien Tixeuil 0.563830
Michèle Sebag 0.388321
Johanne Cohen 0.260465
Albert Cohen 0.510345
Wendy E. Mackay 0.210870
Philippe Caillou 0.558140
Alain Denise 0.288889
Jean-Daniel Fekete 0.414465
Emmanuel Pietriga 0.233333
Yann Ponty 0.559091
Marc Schoenauer 0.420863
Franck Cappello 0.458537
Caroline Appert 0.236957
Michel Beaudouin-Lafon 0.282716
Wendy Mackay 0.234043
Anne Auger 0.530380
Evelyne Lutton 0.227027
Pierre Dragicevic 0.308642
Ioana Manolescu 0.443902
Nikolaus Hansen 0.613580
Nicolas Bredeche 0.491176
Olivier Teytaud 0.494340
François Goasdoué 0.311321
Nathalie Pernelle 0.408824
Fatiha Saïs 0.473171
Sarah Cohen-Boulakia 0.487879
Claude Marché 0.372340
Chantal Reynaud 0.470000
Olivier Chapuis 0.317308
Steven Martin 0.346154
Fatiha Zaidi 0.375000
Balázs Kégl 0.228947
Paola Tubaro 0.497436
Raymond Ros 0.535294
Cyril Furtlehner 0.571795
Anastasia Bezerianos 0.311940
Sylvie Boldo 0.468571
Guillaume Melquiond 0.506061
Marc Baboulin 0.480000
Dimo Brockhoff 0.605128
Nathann Cohen 0.446341
Petra Isenberg 0.409346
Tobias Isenberg 0.570940
Loïc Paulevé 0.769048
Céline Gicquel 0.660526
Isabelle Guyon 0.577528
Guillaume Charpiat 0.396774
Lonni Besançon 0.427273
dtype: float64
3_2D__2_nD_3_2D_neighbors_articles_authors_dim
2_nD__neighbors_articles_authors \
Sébastien Tixeuil 0.680851
Michèle Sebag 0.383942
Johanne Cohen 0.272093
Albert Cohen 0.529310
Wendy E. Mackay 0.200000
Philippe Caillou 0.674419
Alain Denise 0.297222
Jean-Daniel Fekete 0.437107
Emmanuel Pietriga 0.260317
Yann Ponty 0.586364
Marc Schoenauer 0.443165
Franck Cappello 0.480488
Caroline Appert 0.273913
Michel Beaudouin-Lafon 0.297531
Wendy Mackay 0.302128
Anne Auger 0.591139
Evelyne Lutton 0.272973
Pierre Dragicevic 0.317284
Ioana Manolescu 0.507317
Nikolaus Hansen 0.686420
Nicolas Bredeche 0.544118
Olivier Teytaud 0.567925
François Goasdoué 0.430189
Nathalie Pernelle 0.482353
Fatiha Saïs 0.507317
Sarah Cohen-Boulakia 0.581818
Claude Marché 0.465957
Chantal Reynaud 0.513333
Olivier Chapuis 0.384615
Steven Martin 0.356410
Fatiha Zaidi 0.390625
Balázs Kégl 0.286842
Paola Tubaro 0.602564
Raymond Ros 0.655882
Cyril Furtlehner 0.666667
Anastasia Bezerianos 0.319403
Sylvie Boldo 0.488571
Guillaume Melquiond 0.472727
Marc Baboulin 0.540000
Dimo Brockhoff 0.610256
Nathann Cohen 0.380488
Petra Isenberg 0.479439
Tobias Isenberg 0.536752
Loïc Paulevé 0.726190
Céline Gicquel 0.818421
Isabelle Guyon 0.623596
Guillaume Charpiat 0.522581
Lonni Besançon 0.487879
3_2D__neighbors_articles_authors scdec
Sébastien Tixeuil 0.563830 -0.117021
Michèle Sebag 0.388321 0.004380
Johanne Cohen 0.260465 -0.011628
Albert Cohen 0.510345 -0.018966
Wendy E. Mackay 0.210870 0.010870
Philippe Caillou 0.558140 -0.116279
Alain Denise 0.288889 -0.008333
Jean-Daniel Fekete 0.414465 -0.022642
Emmanuel Pietriga 0.233333 -0.026984
Yann Ponty 0.559091 -0.027273
Marc Schoenauer 0.420863 -0.022302
Franck Cappello 0.458537 -0.021951
Caroline Appert 0.236957 -0.036957
Michel Beaudouin-Lafon 0.282716 -0.014815
Wendy Mackay 0.234043 -0.068085
Anne Auger 0.530380 -0.060759
Evelyne Lutton 0.227027 -0.045946
Pierre Dragicevic 0.308642 -0.008642
Ioana Manolescu 0.443902 -0.063415
Nikolaus Hansen 0.613580 -0.072840
Nicolas Bredeche 0.491176 -0.052941
Olivier Teytaud 0.494340 -0.073585
François Goasdoué 0.311321 -0.118868
Nathalie Pernelle 0.408824 -0.073529
Fatiha Saïs 0.473171 -0.034146
Sarah Cohen-Boulakia 0.487879 -0.093939
Claude Marché 0.372340 -0.093617
Chantal Reynaud 0.470000 -0.043333
Olivier Chapuis 0.317308 -0.067308
Steven Martin 0.346154 -0.010256
Fatiha Zaidi 0.375000 -0.015625
Balázs Kégl 0.228947 -0.057895
Paola Tubaro 0.497436 -0.105128
Raymond Ros 0.535294 -0.120588
Cyril Furtlehner 0.571795 -0.094872
Anastasia Bezerianos 0.311940 -0.007463
Sylvie Boldo 0.468571 -0.020000
Guillaume Melquiond 0.506061 0.033333
Marc Baboulin 0.480000 -0.060000
Dimo Brockhoff 0.605128 -0.005128
Nathann Cohen 0.446341 0.065854
Petra Isenberg 0.409346 -0.070093
Tobias Isenberg 0.570940 0.034188
Loïc Paulevé 0.769048 0.042857
Céline Gicquel 0.660526 -0.157895
Isabelle Guyon 0.577528 -0.046067
Guillaume Charpiat 0.396774 -0.125806
Lonni Besançon 0.427273 -0.060606
3_projection_2D scores:
None
Nothing in cache, initial Fitting with min_cluster_size=15 Found 84 clusters in 0.2575946639990434s
Max Fitting with min_cluster_size=30 Found 47 clusters in 0.1043320050011971s
Max Fitting with min_cluster_size=60 Found 23 clusters in 0.10111476800011587s
Max Fitting with min_cluster_size=120 Found 11 clusters in 0.09537671500220313s
Max Fitting with min_cluster_size=240 Found 3 clusters in 0.09179714699712349s
Midpoint Fitting with min_cluster_size=180 Found 9 clusters in 0.09135648099982063s
Midpoint Fitting with min_cluster_size=210 Found 3 clusters in 0.09339748499769485s
Midpoint Fitting with min_cluster_size=195 Found 3 clusters in 0.09393366999938735s
Midpoint Fitting with min_cluster_size=187 Found 3 clusters in 0.09419226499812794s
Re-Fitting with min_cluster_size=180 Found 9 clusters in 0.09104692099936074s
Clusters cached: [3, 3, 3, 3, 9, 11, 23, 47, 84]
Nothing in cache, initial Fitting with min_cluster_size=15 Found 84 clusters in 0.09817058000044199s
Max Fitting with min_cluster_size=30 Found 47 clusters in 0.09374955499879434s
Max Fitting with min_cluster_size=60 Found 23 clusters in 0.09158697800012305s
Midpoint Fitting with min_cluster_size=45 Found 31 clusters in 0.09903591299735126s
Midpoint Fitting with min_cluster_size=52 Found 29 clusters in 0.0987786400000914s
No need Re-Fitting with min_cluster_size=52
Clusters cached: [23, 29, 31, 47, 84]
================================
Run params : {'4_clus__rkey': 'hdbscan', '4_clus__rbase_factor': 3}
--------------------------------
Scoring params : {'4_clus__siter_stab': 2, '4_clus__sremove_stab': [0, 0.01, 0.03, 0.1, 0.25], '4_clus__smetric': 'euclidean', '4_clus__srandom_state': None}
================================
----------- Scores -----------
4_clus__nb_clust_0 : 8.0000
4_clus__silhouette_0 : 0.1605
4_clus__avg_word_couv_0 : 0.4438
4_clus__med_word_couv_0 : 0.4656
4_clus__avg_word_couv_minus_0 : 0.4143
4_clus__big_small_ratio_0 : 6.2094
4_clus__stab_clus_0 : 0.2333
4_clus__nb_clust_1 : 24.0000
4_clus__silhouette_1 : 0.2479
4_clus__avg_word_couv_1 : 0.6254
4_clus__med_word_couv_1 : 0.6048
4_clus__avg_word_couv_minus_1 : 0.6021
4_clus__big_small_ratio_1 : 16.8947
4_clus__stab_clus_1 : 0.1724
4_clus__avg_stab_avg : 0.2029
4_clus__avg_couv_avg : 0.5346
4_clus__clu_score : 0.3687
--------- Desc Scores ---------
4_clus__clus_eval_pos_0_det
0 visualizations, interaction techniques : s 538...
1 quantitative methods, bioinformatics : s 455 s...
2 ontology, documents : s 404 stb 0.00 + 0.36 - ...
3 networking internet architecture, cluster comp...
4 matrix, architectures : s 307 stb 0.10 + 0.49 ...
5 discrete mathematics, combinatorics : s 295 st...
6 logic science, semantics : s 282 stb 0.10 + 0....
7 image, french : s 211 stb 0.20 + 0.51 - 0.04
8 fluid, operations : s 191 stb 0.00 + 0.53 - 0.04
dtype: object
VALUE : 0.4438
4_clus__clus_eval_pos_1_det
0 verification, testing : s 282 stb 0.00 + 0.56 ...
1 displays, interfaces : s 277 stb 0.10 + 0.41 -...
2 networking internet, services : s 217 stb 0.00...
3 humanities social sciences, social networks : ...
4 query : s 131 stb 0.10 + 0.81 - 0.02
5 finite, trees : s 131 stb 0.20 + 0.49 - 0.06
6 vertices, minimum : s 129 stb 0.90 + 0.69 - 0.02
7 biology, genes : s 127 stb 0.00 + 0.47 - 0.03
8 secondary structure, protein : s 124 stb 0.00 ...
9 compiler, optimizations : s 120 stb 0.00 + 0.5...
10 dynamics, mechanics : s 120 stb 0.00 + 0.65 - ...
11 fault, cloud : s 118 stb 0.00 + 0.65 - 0.03
12 neural networks, reinforcement learning : s 11...
13 ontologies, linked : s 113 stb 0.00 + 0.58 - 0.03
14 inference, traffic : s 111 stb 0.60 + 0.51 - 0.03
15 information visualization, cognitive science :...
16 multi armed, robotics : s 93 stb 0.00 + 0.61 -...
17 natural language, speech : s 91 stb 0.00 + 0.5...
18 creative, movement : s 88 stb 0.00 + 0.60 - 0.01
19 image processing, vision pattern recognition :...
20 monte carlo, games : s 76 stb 0.90 + 0.95 - 0.02
21 document, semantic annotations : s 76 stb 0.00...
22 floating point, mathematical : s 70 stb 0.10 +...
23 evolution strategies, noisy optimization : s 6...
24 linear systems, matrices : s 61 stb 0.00 + 0.7...
25 stochastic, mixed integer linear : s 60 stb 0....
26 multi objective, surrogate model : s 59 stb 0....
27 benchmarking : s 58 stb 1.00 + 0.84 - 0.01
28 visual analytics : s 57 stb 0.00 + 0.82 - 0.01
dtype: object
VALUE : 0.6254
--------- Raw Scores ---------
4_clustering scores:
None
['2_nD__neighbors_articles_authors', '3_2D__neighbors_articles_authors', '4_clus__clu_score']
================================
Run params : {}
--------------------------------
Scoring params : {'6_pst__sname_list': ['2_nD__neighbors_articles_authors', '3_2D__neighbors_articles_authors', '4_clus__clu_score']}
================================
----------- Scores -----------
6_pst__final_score : 0.4263
--------- Desc Scores ---------
6_pst__final_score_det
0 2_nD__neighbors_articles_authors : 0.48
1 3_2D__neighbors_articles_authors : 0.43
2 4_clus__clu_score : 0.37
dtype: object
VALUE : 0.4263
--------- Raw Scores ---------
6_post scores:
None
Finished running step!
Running experiment for parameters:
{'id': 'ce1c0d334fc75a6570a5', 'robustseed': 0, 'authors__filter_min_score': 4, 'filter_min_score': 6, 'projection_nD': {'key': 'lsa', 'num_dims': 50, 'extra_param': True}, 'projection_2D': {'key': 'umap', 'n_neighbors': 10, 'min_dist': 0.25, 'metric': 'euclidean'}}
Matrices of type mat for articles, authors, teams, labs, words already exist. Skipping creation.
Matrices of type lsa for articles, authors, teams, labs, words already exist. Skipping creation.
================================
Run params : {'2_nD__rkey': 'lsa', '2_nD__rnum_dims': 50, '2_nD__rnormalize': True, '2_nD__rextra_param': True}
--------------------------------
Scoring params : {'2_nD__smin_score': 30, '2_nD__srecompute': True, '2_nD__ssample_size': None, '2_nD__sn_neighbors': 10, '2_nD__srandom_state': 42}
================================
----------- Scores -----------
2_nD__neighbors_articles_authors : 0.4779
--------- Desc Scores ---------
2_nD__neighbors_articles_authors_det
0 Céline Gicquel : 0.82
1 Loïc Paulevé : 0.73
2 Nikolaus Hansen : 0.69
3 Sébastien Tixeuil : 0.68
4 Philippe Caillou : 0.67
5 Cyril Furtlehner : 0.67
6 Raymond Ros : 0.66
7 Isabelle Guyon : 0.62
8 Dimo Brockhoff : 0.61
9 Paola Tubaro : 0.60
10 Anne Auger : 0.59
11 Yann Ponty : 0.59
12 Sarah Cohen-Boulakia : 0.58
13 Olivier Teytaud : 0.57
14 Nicolas Bredeche : 0.54
15 Marc Baboulin : 0.54
16 Tobias Isenberg : 0.54
17 Albert Cohen : 0.53
18 Guillaume Charpiat : 0.52
19 Chantal Reynaud : 0.51
20 Fatiha Saïs : 0.51
21 Ioana Manolescu : 0.51
22 Sylvie Boldo : 0.49
23 Lonni Besançon : 0.49
24 Nathalie Pernelle : 0.48
25 Franck Cappello : 0.48
26 Petra Isenberg : 0.48
27 Guillaume Melquiond : 0.47
28 Claude Marché : 0.47
29 Marc Schoenauer : 0.44
30 Jean-Daniel Fekete : 0.44
31 François Goasdoué : 0.43
32 Fatiha Zaidi : 0.39
33 Olivier Chapuis : 0.38
34 Michèle Sebag : 0.38
35 Nathann Cohen : 0.38
36 Steven Martin : 0.36
37 Anastasia Bezerianos : 0.32
38 Pierre Dragicevic : 0.32
39 Wendy Mackay : 0.30
40 Michel Beaudouin-Lafon : 0.30
41 Alain Denise : 0.30
42 Balázs Kégl : 0.29
43 Caroline Appert : 0.27
44 Evelyne Lutton : 0.27
45 Johanne Cohen : 0.27
46 Emmanuel Pietriga : 0.26
47 Wendy E. Mackay : 0.20
Name: 0, dtype: object
VALUE : 0.4779
--------- Raw Scores ---------
2_nD__neighbors_articles_authors
Sébastien Tixeuil 0.680851
Michèle Sebag 0.383942
Johanne Cohen 0.272093
Albert Cohen 0.529310
Wendy E. Mackay 0.200000
Philippe Caillou 0.674419
Alain Denise 0.297222
Jean-Daniel Fekete 0.437107
Emmanuel Pietriga 0.260317
Yann Ponty 0.586364
Marc Schoenauer 0.443165
Franck Cappello 0.480488
Caroline Appert 0.273913
Michel Beaudouin-Lafon 0.297531
Wendy Mackay 0.302128
Anne Auger 0.591139
Evelyne Lutton 0.272973
Pierre Dragicevic 0.317284
Ioana Manolescu 0.507317
Nikolaus Hansen 0.686420
Nicolas Bredeche 0.544118
Olivier Teytaud 0.567925
François Goasdoué 0.430189
Nathalie Pernelle 0.482353
Fatiha Saïs 0.507317
Sarah Cohen-Boulakia 0.581818
Claude Marché 0.465957
Chantal Reynaud 0.513333
Olivier Chapuis 0.384615
Steven Martin 0.356410
Fatiha Zaidi 0.390625
Balázs Kégl 0.286842
Paola Tubaro 0.602564
Raymond Ros 0.655882
Cyril Furtlehner 0.666667
Anastasia Bezerianos 0.319403
Sylvie Boldo 0.488571
Guillaume Melquiond 0.472727
Marc Baboulin 0.540000
Dimo Brockhoff 0.610256
Nathann Cohen 0.380488
Petra Isenberg 0.479439
Tobias Isenberg 0.536752
Loïc Paulevé 0.726190
Céline Gicquel 0.818421
Isabelle Guyon 0.623596
Guillaume Charpiat 0.522581
Lonni Besançon 0.487879
Name: 0, dtype: float64
================================
Run params : {}
--------------------------------
Scoring params : {}
================================
----------- Scores -----------
custom_score_lsa : 10.0000
--------- Desc Scores ---------
--------- Raw Scores ---------
custom scores:
None
================================
Run params : {'2_nD__rkey': 'lsa', '2_nD__rnum_dims': 50, '2_nD__rnormalize': True, '2_nD__rextra_param': True}
--------------------------------
Scoring params : {'2_nD__smin_score': 30, '2_nD__srecompute': True, '2_nD__ssample_size': None, '2_nD__sn_neighbors': 10, '2_nD__srandom_state': 42, '2_nD__sfactor': 10}
================================
----------- Scores -----------
2_nD__neighbors_articles_authors : 0.4779
2_nD__custom_score_lsa : 10.0000
--------- Desc Scores ---------
2_nD__neighbors_articles_authors_det
0 Céline Gicquel : 0.82
1 Loïc Paulevé : 0.73
2 Nikolaus Hansen : 0.69
3 Sébastien Tixeuil : 0.68
4 Philippe Caillou : 0.67
5 Cyril Furtlehner : 0.67
6 Raymond Ros : 0.66
7 Isabelle Guyon : 0.62
8 Dimo Brockhoff : 0.61
9 Paola Tubaro : 0.60
10 Anne Auger : 0.59
11 Yann Ponty : 0.59
12 Sarah Cohen-Boulakia : 0.58
13 Olivier Teytaud : 0.57
14 Nicolas Bredeche : 0.54
15 Marc Baboulin : 0.54
16 Tobias Isenberg : 0.54
17 Albert Cohen : 0.53
18 Guillaume Charpiat : 0.52
19 Chantal Reynaud : 0.51
20 Fatiha Saïs : 0.51
21 Ioana Manolescu : 0.51
22 Sylvie Boldo : 0.49
23 Lonni Besançon : 0.49
24 Nathalie Pernelle : 0.48
25 Franck Cappello : 0.48
26 Petra Isenberg : 0.48
27 Guillaume Melquiond : 0.47
28 Claude Marché : 0.47
29 Marc Schoenauer : 0.44
30 Jean-Daniel Fekete : 0.44
31 François Goasdoué : 0.43
32 Fatiha Zaidi : 0.39
33 Olivier Chapuis : 0.38
34 Michèle Sebag : 0.38
35 Nathann Cohen : 0.38
36 Steven Martin : 0.36
37 Anastasia Bezerianos : 0.32
38 Pierre Dragicevic : 0.32
39 Wendy Mackay : 0.30
40 Michel Beaudouin-Lafon : 0.30
41 Alain Denise : 0.30
42 Balázs Kégl : 0.29
43 Caroline Appert : 0.27
44 Evelyne Lutton : 0.27
45 Johanne Cohen : 0.27
46 Emmanuel Pietriga : 0.26
47 Wendy E. Mackay : 0.20
Name: 0, dtype: object
VALUE : 0.4779
--------- Raw Scores ---------
2_nD__neighbors_articles_authors
Sébastien Tixeuil 0.680851
Michèle Sebag 0.383942
Johanne Cohen 0.272093
Albert Cohen 0.529310
Wendy E. Mackay 0.200000
Philippe Caillou 0.674419
Alain Denise 0.297222
Jean-Daniel Fekete 0.437107
Emmanuel Pietriga 0.260317
Yann Ponty 0.586364
Marc Schoenauer 0.443165
Franck Cappello 0.480488
Caroline Appert 0.273913
Michel Beaudouin-Lafon 0.297531
Wendy Mackay 0.302128
Anne Auger 0.591139
Evelyne Lutton 0.272973
Pierre Dragicevic 0.317284
Ioana Manolescu 0.507317
Nikolaus Hansen 0.686420
Nicolas Bredeche 0.544118
Olivier Teytaud 0.567925
François Goasdoué 0.430189
Nathalie Pernelle 0.482353
Fatiha Saïs 0.507317
Sarah Cohen-Boulakia 0.581818
Claude Marché 0.465957
Chantal Reynaud 0.513333
Olivier Chapuis 0.384615
Steven Martin 0.356410
Fatiha Zaidi 0.390625
Balázs Kégl 0.286842
Paola Tubaro 0.602564
Raymond Ros 0.655882
Cyril Furtlehner 0.666667
Anastasia Bezerianos 0.319403
Sylvie Boldo 0.488571
Guillaume Melquiond 0.472727
Marc Baboulin 0.540000
Dimo Brockhoff 0.610256
Nathann Cohen 0.380488
Petra Isenberg 0.479439
Tobias Isenberg 0.536752
Loïc Paulevé 0.726190
Céline Gicquel 0.818421
Isabelle Guyon 0.623596
Guillaume Charpiat 0.522581
Lonni Besançon 0.487879
Name: 0, dtype: float64
2_projection_nD scores:
None
umap matrices generated.
================================
Run params : {'3_2D__rkey': 'umap', '3_2D__rmetric': 'euclidean', '3_2D__rn_neighbors': 10, '3_2D__rmin_dist': 0.25, '3_2D__rinit': 'random', '3_2D__rlearning_rate': 1.0, '3_2D__rn_epochs': None, '3_2D__rrandom_state': None}
--------------------------------
Scoring params : {'3_2D__smin_score': 30, '3_2D__srecompute': True, '3_2D__ssample_size': None, '3_2D__sn_neighbors': 10, '3_2D__srandom_state': 42}
================================
----------- Scores -----------
3_2D__neighbors_articles_authors : 0.4255
3_2D__2_nD_3_2D_neighbors_articles_authors_dim_mean : -0.0524
3_2D__2_nD_3_2D_neighbors_articles_authors_dim_median : -0.0467
--------- Desc Scores ---------
3_2D__neighbors_articles_authors_det
0 Loïc Paulevé : 0.73
1 Céline Gicquel : 0.67
2 Isabelle Guyon : 0.59
3 Philippe Caillou : 0.59
4 Nikolaus Hansen : 0.58
5 Dimo Brockhoff : 0.58
6 Sébastien Tixeuil : 0.57
7 Yann Ponty : 0.57
8 Cyril Furtlehner : 0.56
9 Tobias Isenberg : 0.55
10 Raymond Ros : 0.55
11 Nicolas Bredeche : 0.52
12 Fatiha Saïs : 0.50
13 Anne Auger : 0.49
14 Marc Baboulin : 0.48
15 Paola Tubaro : 0.48
16 Guillaume Melquiond : 0.48
17 Olivier Teytaud : 0.47
18 Albert Cohen : 0.47
19 Sylvie Boldo : 0.47
20 Franck Cappello : 0.45
21 Sarah Cohen-Boulakia : 0.44
22 Lonni Besançon : 0.44
23 Ioana Manolescu : 0.43
24 Chantal Reynaud : 0.43
25 Marc Schoenauer : 0.40
26 Nathann Cohen : 0.40
27 Guillaume Charpiat : 0.40
28 Claude Marché : 0.40
29 Jean-Daniel Fekete : 0.39
30 Fatiha Zaidi : 0.38
31 Petra Isenberg : 0.38
32 Nathalie Pernelle : 0.38
33 Steven Martin : 0.37
34 Michèle Sebag : 0.36
35 François Goasdoué : 0.33
36 Michel Beaudouin-Lafon : 0.31
37 Olivier Chapuis : 0.31
38 Pierre Dragicevic : 0.30
39 Anastasia Bezerianos : 0.30
40 Johanne Cohen : 0.27
41 Alain Denise : 0.26
42 Caroline Appert : 0.26
43 Emmanuel Pietriga : 0.24
44 Wendy Mackay : 0.24
45 Balázs Kégl : 0.23
46 Evelyne Lutton : 0.22
47 Wendy E. Mackay : 0.21
dtype: object
VALUE : 0.4255
3_2D__2_nD_3_2D_neighbors_articles_authors_dim_det
0 Sarah Cohen-Boulakia : -0.15
1 Céline Gicquel : -0.14
2 Guillaume Charpiat : -0.12
3 Paola Tubaro : -0.12
4 Sébastien Tixeuil : -0.11
5 Raymond Ros : -0.11
6 Nikolaus Hansen : -0.11
7 Anne Auger : -0.10
8 François Goasdoué : -0.10
9 Cyril Furtlehner : -0.10
10 Nathalie Pernelle : -0.10
11 Olivier Teytaud : -0.10
12 Petra Isenberg : -0.10
13 Chantal Reynaud : -0.09
14 Philippe Caillou : -0.08
15 Ioana Manolescu : -0.08
16 Olivier Chapuis : -0.08
17 Claude Marché : -0.07
18 Balázs Kégl : -0.06
19 Wendy Mackay : -0.06
20 Albert Cohen : -0.06
21 Marc Baboulin : -0.06
22 Lonni Besançon : -0.05
23 Evelyne Lutton : -0.05
24 Jean-Daniel Fekete : -0.04
25 Marc Schoenauer : -0.04
26 Alain Denise : -0.04
27 Dimo Brockhoff : -0.03
28 Isabelle Guyon : -0.03
29 Michèle Sebag : -0.03
30 Franck Cappello : -0.03
31 Anastasia Bezerianos : -0.02
32 Nicolas Bredeche : -0.02
33 Yann Ponty : -0.02
34 Sylvie Boldo : -0.02
35 Emmanuel Pietriga : -0.02
36 Caroline Appert : -0.02
37 Pierre Dragicevic : -0.02
38 Fatiha Saïs : -0.01
39 Fatiha Zaidi : -0.01
40 Johanne Cohen : -0.00
41 Loïc Paulevé : 0.00
42 Guillaume Melquiond : 0.01
43 Wendy E. Mackay : 0.01
44 Michel Beaudouin-Lafon : 0.01
45 Tobias Isenberg : 0.01
46 Steven Martin : 0.02
47 Nathann Cohen : 0.02
dtype: object
VALUE : -0.0524
--------- Raw Scores ---------
3_2D__neighbors_articles_authors
Sébastien Tixeuil 0.565957
Michèle Sebag 0.356934
Johanne Cohen 0.267442
Albert Cohen 0.470690
Wendy E. Mackay 0.210870
Philippe Caillou 0.593023
Alain Denise 0.261111
Jean-Daniel Fekete 0.392453
Emmanuel Pietriga 0.242857
Yann Ponty 0.565152
Marc Schoenauer 0.402158
Franck Cappello 0.453659
Caroline Appert 0.256522
Michel Beaudouin-Lafon 0.308642
Wendy Mackay 0.242553
Anne Auger 0.487342
Evelyne Lutton 0.224324
Pierre Dragicevic 0.300000
Ioana Manolescu 0.426829
Nikolaus Hansen 0.579012
Nicolas Bredeche 0.520588
Olivier Teytaud 0.470755
François Goasdoué 0.326415
Nathalie Pernelle 0.382353
Fatiha Saïs 0.495122
Sarah Cohen-Boulakia 0.436364
Claude Marché 0.397872
Chantal Reynaud 0.426667
Olivier Chapuis 0.307692
Steven Martin 0.374359
Fatiha Zaidi 0.384375
Balázs Kégl 0.226316
Paola Tubaro 0.482051
Raymond Ros 0.547059
Cyril Furtlehner 0.564103
Anastasia Bezerianos 0.295522
Sylvie Boldo 0.468571
Guillaume Melquiond 0.481818
Marc Baboulin 0.484444
Dimo Brockhoff 0.576923
Nathann Cohen 0.400000
Petra Isenberg 0.384112
Tobias Isenberg 0.550427
Loïc Paulevé 0.726190
Céline Gicquel 0.673684
Isabelle Guyon 0.594382
Guillaume Charpiat 0.400000
Lonni Besançon 0.436364
dtype: float64
3_2D__2_nD_3_2D_neighbors_articles_authors_dim
2_nD__neighbors_articles_authors \
Sébastien Tixeuil 0.680851
Michèle Sebag 0.383942
Johanne Cohen 0.272093
Albert Cohen 0.529310
Wendy E. Mackay 0.200000
Philippe Caillou 0.674419
Alain Denise 0.297222
Jean-Daniel Fekete 0.437107
Emmanuel Pietriga 0.260317
Yann Ponty 0.586364
Marc Schoenauer 0.443165
Franck Cappello 0.480488
Caroline Appert 0.273913
Michel Beaudouin-Lafon 0.297531
Wendy Mackay 0.302128
Anne Auger 0.591139
Evelyne Lutton 0.272973
Pierre Dragicevic 0.317284
Ioana Manolescu 0.507317
Nikolaus Hansen 0.686420
Nicolas Bredeche 0.544118
Olivier Teytaud 0.567925
François Goasdoué 0.430189
Nathalie Pernelle 0.482353
Fatiha Saïs 0.507317
Sarah Cohen-Boulakia 0.581818
Claude Marché 0.465957
Chantal Reynaud 0.513333
Olivier Chapuis 0.384615
Steven Martin 0.356410
Fatiha Zaidi 0.390625
Balázs Kégl 0.286842
Paola Tubaro 0.602564
Raymond Ros 0.655882
Cyril Furtlehner 0.666667
Anastasia Bezerianos 0.319403
Sylvie Boldo 0.488571
Guillaume Melquiond 0.472727
Marc Baboulin 0.540000
Dimo Brockhoff 0.610256
Nathann Cohen 0.380488
Petra Isenberg 0.479439
Tobias Isenberg 0.536752
Loïc Paulevé 0.726190
Céline Gicquel 0.818421
Isabelle Guyon 0.623596
Guillaume Charpiat 0.522581
Lonni Besançon 0.487879
3_2D__neighbors_articles_authors scdec
Sébastien Tixeuil 0.565957 -0.114894
Michèle Sebag 0.356934 -0.027007
Johanne Cohen 0.267442 -0.004651
Albert Cohen 0.470690 -0.058621
Wendy E. Mackay 0.210870 0.010870
Philippe Caillou 0.593023 -0.081395
Alain Denise 0.261111 -0.036111
Jean-Daniel Fekete 0.392453 -0.044654
Emmanuel Pietriga 0.242857 -0.017460
Yann Ponty 0.565152 -0.021212
Marc Schoenauer 0.402158 -0.041007
Franck Cappello 0.453659 -0.026829
Caroline Appert 0.256522 -0.017391
Michel Beaudouin-Lafon 0.308642 0.011111
Wendy Mackay 0.242553 -0.059574
Anne Auger 0.487342 -0.103797
Evelyne Lutton 0.224324 -0.048649
Pierre Dragicevic 0.300000 -0.017284
Ioana Manolescu 0.426829 -0.080488
Nikolaus Hansen 0.579012 -0.107407
Nicolas Bredeche 0.520588 -0.023529
Olivier Teytaud 0.470755 -0.097170
François Goasdoué 0.326415 -0.103774
Nathalie Pernelle 0.382353 -0.100000
Fatiha Saïs 0.495122 -0.012195
Sarah Cohen-Boulakia 0.436364 -0.145455
Claude Marché 0.397872 -0.068085
Chantal Reynaud 0.426667 -0.086667
Olivier Chapuis 0.307692 -0.076923
Steven Martin 0.374359 0.017949
Fatiha Zaidi 0.384375 -0.006250
Balázs Kégl 0.226316 -0.060526
Paola Tubaro 0.482051 -0.120513
Raymond Ros 0.547059 -0.108824
Cyril Furtlehner 0.564103 -0.102564
Anastasia Bezerianos 0.295522 -0.023881
Sylvie Boldo 0.468571 -0.020000
Guillaume Melquiond 0.481818 0.009091
Marc Baboulin 0.484444 -0.055556
Dimo Brockhoff 0.576923 -0.033333
Nathann Cohen 0.400000 0.019512
Petra Isenberg 0.384112 -0.095327
Tobias Isenberg 0.550427 0.013675
Loïc Paulevé 0.726190 0.000000
Céline Gicquel 0.673684 -0.144737
Isabelle Guyon 0.594382 -0.029213
Guillaume Charpiat 0.400000 -0.122581
Lonni Besançon 0.436364 -0.051515
3_projection_2D scores:
None
Nothing in cache, initial Fitting with min_cluster_size=15 Found 70 clusters in 0.271356138997362s
Max Fitting with min_cluster_size=30 Found 46 clusters in 0.10519513500184985s
Max Fitting with min_cluster_size=60 Found 22 clusters in 0.10125969300133875s
Max Fitting with min_cluster_size=120 Found 3 clusters in 0.10196087800068199s
Midpoint Fitting with min_cluster_size=90 Found 10 clusters in 0.0980560729985882s
Midpoint Fitting with min_cluster_size=105 Found 9 clusters in 0.09737261600093916s
Midpoint Fitting with min_cluster_size=112 Found 7 clusters in 0.10034898000230896s
Re-Fitting with min_cluster_size=105 Found 9 clusters in 0.09951053599797888s
Clusters cached: [3, 7, 9, 10, 22, 46, 70]
Nothing in cache, initial Fitting with min_cluster_size=15 Found 70 clusters in 0.10536656699696323s
Max Fitting with min_cluster_size=30 Found 46 clusters in 0.10532998599956045s
Max Fitting with min_cluster_size=60 Found 22 clusters in 0.10093085799962864s
Midpoint Fitting with min_cluster_size=45 Found 29 clusters in 0.10149668499798281s
Midpoint Fitting with min_cluster_size=52 Found 25 clusters in 0.1021060840030259s
No need Re-Fitting with min_cluster_size=52
Clusters cached: [22, 25, 29, 46, 70]
================================
Run params : {'4_clus__rkey': 'hdbscan', '4_clus__rbase_factor': 3}
--------------------------------
Scoring params : {'4_clus__siter_stab': 2, '4_clus__sremove_stab': [0, 0.01, 0.03, 0.1, 0.25], '4_clus__smetric': 'euclidean', '4_clus__srandom_state': None}
================================
----------- Scores -----------
4_clus__nb_clust_0 : 8.0000
4_clus__silhouette_0 : -0.0544
4_clus__avg_word_couv_0 : 0.5173
4_clus__med_word_couv_0 : 0.5374
4_clus__avg_word_couv_minus_0 : 0.4884
4_clus__big_small_ratio_0 : 14.3246
4_clus__stab_clus_0 : 0.0444
4_clus__nb_clust_1 : 24.0000
4_clus__silhouette_1 : 0.0569
4_clus__avg_word_couv_1 : 0.6466
4_clus__med_word_couv_1 : 0.6961
4_clus__avg_word_couv_minus_1 : 0.6206
4_clus__big_small_ratio_1 : 26.2830
4_clus__stab_clus_1 : 0.0480
4_clus__avg_stab_avg : 0.0462
4_clus__avg_couv_avg : 0.5819
4_clus__clu_score : 0.3141
--------- Desc Scores ---------
4_clus__clus_eval_pos_0_det
0 query, networking internet architecture : s 11...
1 visualizations, interfaces : s 586 stb 0.30 + ...
2 benchmarking, multi objective : s 147 stb 0.00...
3 logic science, semantics : s 138 stb 0.00 + 0....
4 vertices, minimum : s 132 stb 0.10 + 0.68 - 0.02
5 humanities social sciences, social networks : ...
6 fluid, dynamics : s 125 stb 0.00 + 0.67 - 0.05
7 combinatorics, automata : s 115 stb 0.00 + 0.5...
8 biological, metabolic : s 114 stb 0.00 + 0.57 ...
dtype: object
VALUE : 0.5173
4_clus__clus_eval_pos_1_det
0 interaction techniques, physical : s 529 stb 0...
1 ontology, documents : s 273 stb 0.20 + 0.41 - ...
2 networking internet, reinforcement learning : ...
3 verification, calculus : s 138 stb 0.00 + 0.66...
4 discrete mathematics, vertex : s 132 stb 0.00 ...
5 humanities social, community : s 129 stb 0.00 ...
6 numerical simulations, mechanics : s 125 stb 0...
7 trees, finite : s 115 stb 0.00 + 0.52 - 0.06
8 genes, metabolism : s 114 stb 0.00 + 0.40 - 0.01
9 architectures, programming languages : s 102 s...
10 queries : s 98 stb 0.00 + 0.76 - 0.02
11 noisy optimization, gradient : s 82 stb 0.00 +...
12 secondary structure, sequence : s 81 stb 0.00 ...
13 image, motion : s 77 stb 0.00 + 0.87 - 0.03
14 testing, model checking : s 74 stb 0.00 + 0.77...
15 operations, bounds : s 72 stb 0.00 + 0.75 - 0.07
16 protein, molecular : s 71 stb 0.00 + 0.83 - 0.02
17 services, cloud : s 71 stb 0.10 + 0.86 - 0.04
18 testbed, black : s 71 stb 0.00 + 0.89 - 0.03
19 multi objective optimization, planning : s 68 ...
20 stabilizing, population protocols : s 62 stb 0...
21 linear systems, matrices : s 60 stb 0.40 + 0.7...
22 floating point : s 55 stb 0.00 + 0.80 - 0.00
23 belief propagation, traffic : s 53 stb 0.00 + ...
24 visual analytics : s 53 stb 0.00 + 0.87 - 0.01
dtype: object
VALUE : 0.6466
--------- Raw Scores ---------
4_clustering scores:
None
['2_nD__neighbors_articles_authors', '3_2D__neighbors_articles_authors', '4_clus__clu_score']
================================
Run params : {}
--------------------------------
Scoring params : {'6_pst__sname_list': ['2_nD__neighbors_articles_authors', '3_2D__neighbors_articles_authors', '4_clus__clu_score']}
================================
----------- Scores -----------
6_pst__final_score : 0.4058
--------- Desc Scores ---------
6_pst__final_score_det
0 2_nD__neighbors_articles_authors : 0.48
1 3_2D__neighbors_articles_authors : 0.43
2 4_clus__clu_score : 0.31
dtype: object
VALUE : 0.4058
--------- Raw Scores ---------
6_post scores:
None
Finished running step!
Running experiment for parameters:
{'id': 'a588bce09630b37c4111', 'robustseed': 0, 'authors__filter_min_score': 4, 'filter_min_score': 6, 'projection_nD': {'key': 'lsa', 'num_dims': 50, 'extra_param': True}, 'projection_2D': {'key': 'umap', 'n_neighbors': 10, 'min_dist': 0.5, 'metric': 'euclidean'}}
Matrices of type mat for articles, authors, teams, labs, words already exist. Skipping creation.
Matrices of type lsa for articles, authors, teams, labs, words already exist. Skipping creation.
================================
Run params : {'2_nD__rkey': 'lsa', '2_nD__rnum_dims': 50, '2_nD__rnormalize': True, '2_nD__rextra_param': True}
--------------------------------
Scoring params : {'2_nD__smin_score': 30, '2_nD__srecompute': True, '2_nD__ssample_size': None, '2_nD__sn_neighbors': 10, '2_nD__srandom_state': 42}
================================
----------- Scores -----------
2_nD__neighbors_articles_authors : 0.4779
--------- Desc Scores ---------
2_nD__neighbors_articles_authors_det
0 Céline Gicquel : 0.82
1 Loïc Paulevé : 0.73
2 Nikolaus Hansen : 0.69
3 Sébastien Tixeuil : 0.68
4 Philippe Caillou : 0.67
5 Cyril Furtlehner : 0.67
6 Raymond Ros : 0.66
7 Isabelle Guyon : 0.62
8 Dimo Brockhoff : 0.61
9 Paola Tubaro : 0.60
10 Anne Auger : 0.59
11 Yann Ponty : 0.59
12 Sarah Cohen-Boulakia : 0.58
13 Olivier Teytaud : 0.57
14 Nicolas Bredeche : 0.54
15 Marc Baboulin : 0.54
16 Tobias Isenberg : 0.54
17 Albert Cohen : 0.53
18 Guillaume Charpiat : 0.52
19 Chantal Reynaud : 0.51
20 Fatiha Saïs : 0.51
21 Ioana Manolescu : 0.51
22 Sylvie Boldo : 0.49
23 Lonni Besançon : 0.49
24 Nathalie Pernelle : 0.48
25 Franck Cappello : 0.48
26 Petra Isenberg : 0.48
27 Guillaume Melquiond : 0.47
28 Claude Marché : 0.47
29 Marc Schoenauer : 0.44
30 Jean-Daniel Fekete : 0.44
31 François Goasdoué : 0.43
32 Fatiha Zaidi : 0.39
33 Olivier Chapuis : 0.38
34 Michèle Sebag : 0.38
35 Nathann Cohen : 0.38
36 Steven Martin : 0.36
37 Anastasia Bezerianos : 0.32
38 Pierre Dragicevic : 0.32
39 Wendy Mackay : 0.30
40 Michel Beaudouin-Lafon : 0.30
41 Alain Denise : 0.30
42 Balázs Kégl : 0.29
43 Caroline Appert : 0.27
44 Evelyne Lutton : 0.27
45 Johanne Cohen : 0.27
46 Emmanuel Pietriga : 0.26
47 Wendy E. Mackay : 0.20
Name: 0, dtype: object
VALUE : 0.4779
--------- Raw Scores ---------
2_nD__neighbors_articles_authors
Sébastien Tixeuil 0.680851
Michèle Sebag 0.383942
Johanne Cohen 0.272093
Albert Cohen 0.529310
Wendy E. Mackay 0.200000
Philippe Caillou 0.674419
Alain Denise 0.297222
Jean-Daniel Fekete 0.437107
Emmanuel Pietriga 0.260317
Yann Ponty 0.586364
Marc Schoenauer 0.443165
Franck Cappello 0.480488
Caroline Appert 0.273913
Michel Beaudouin-Lafon 0.297531
Wendy Mackay 0.302128
Anne Auger 0.591139
Evelyne Lutton 0.272973
Pierre Dragicevic 0.317284
Ioana Manolescu 0.507317
Nikolaus Hansen 0.686420
Nicolas Bredeche 0.544118
Olivier Teytaud 0.567925
François Goasdoué 0.430189
Nathalie Pernelle 0.482353
Fatiha Saïs 0.507317
Sarah Cohen-Boulakia 0.581818
Claude Marché 0.465957
Chantal Reynaud 0.513333
Olivier Chapuis 0.384615
Steven Martin 0.356410
Fatiha Zaidi 0.390625
Balázs Kégl 0.286842
Paola Tubaro 0.602564
Raymond Ros 0.655882
Cyril Furtlehner 0.666667
Anastasia Bezerianos 0.319403
Sylvie Boldo 0.488571
Guillaume Melquiond 0.472727
Marc Baboulin 0.540000
Dimo Brockhoff 0.610256
Nathann Cohen 0.380488
Petra Isenberg 0.479439
Tobias Isenberg 0.536752
Loïc Paulevé 0.726190
Céline Gicquel 0.818421
Isabelle Guyon 0.623596
Guillaume Charpiat 0.522581
Lonni Besançon 0.487879
Name: 0, dtype: float64
================================
Run params : {}
--------------------------------
Scoring params : {}
================================
----------- Scores -----------
custom_score_lsa : 10.0000
--------- Desc Scores ---------
--------- Raw Scores ---------
custom scores:
None
================================
Run params : {'2_nD__rkey': 'lsa', '2_nD__rnum_dims': 50, '2_nD__rnormalize': True, '2_nD__rextra_param': True}
--------------------------------
Scoring params : {'2_nD__smin_score': 30, '2_nD__srecompute': True, '2_nD__ssample_size': None, '2_nD__sn_neighbors': 10, '2_nD__srandom_state': 42, '2_nD__sfactor': 10}
================================
----------- Scores -----------
2_nD__neighbors_articles_authors : 0.4779
2_nD__custom_score_lsa : 10.0000
--------- Desc Scores ---------
2_nD__neighbors_articles_authors_det
0 Céline Gicquel : 0.82
1 Loïc Paulevé : 0.73
2 Nikolaus Hansen : 0.69
3 Sébastien Tixeuil : 0.68
4 Philippe Caillou : 0.67
5 Cyril Furtlehner : 0.67
6 Raymond Ros : 0.66
7 Isabelle Guyon : 0.62
8 Dimo Brockhoff : 0.61
9 Paola Tubaro : 0.60
10 Anne Auger : 0.59
11 Yann Ponty : 0.59
12 Sarah Cohen-Boulakia : 0.58
13 Olivier Teytaud : 0.57
14 Nicolas Bredeche : 0.54
15 Marc Baboulin : 0.54
16 Tobias Isenberg : 0.54
17 Albert Cohen : 0.53
18 Guillaume Charpiat : 0.52
19 Chantal Reynaud : 0.51
20 Fatiha Saïs : 0.51
21 Ioana Manolescu : 0.51
22 Sylvie Boldo : 0.49
23 Lonni Besançon : 0.49
24 Nathalie Pernelle : 0.48
25 Franck Cappello : 0.48
26 Petra Isenberg : 0.48
27 Guillaume Melquiond : 0.47
28 Claude Marché : 0.47
29 Marc Schoenauer : 0.44
30 Jean-Daniel Fekete : 0.44
31 François Goasdoué : 0.43
32 Fatiha Zaidi : 0.39
33 Olivier Chapuis : 0.38
34 Michèle Sebag : 0.38
35 Nathann Cohen : 0.38
36 Steven Martin : 0.36
37 Anastasia Bezerianos : 0.32
38 Pierre Dragicevic : 0.32
39 Wendy Mackay : 0.30
40 Michel Beaudouin-Lafon : 0.30
41 Alain Denise : 0.30
42 Balázs Kégl : 0.29
43 Caroline Appert : 0.27
44 Evelyne Lutton : 0.27
45 Johanne Cohen : 0.27
46 Emmanuel Pietriga : 0.26
47 Wendy E. Mackay : 0.20
Name: 0, dtype: object
VALUE : 0.4779
--------- Raw Scores ---------
2_nD__neighbors_articles_authors
Sébastien Tixeuil 0.680851
Michèle Sebag 0.383942
Johanne Cohen 0.272093
Albert Cohen 0.529310
Wendy E. Mackay 0.200000
Philippe Caillou 0.674419
Alain Denise 0.297222
Jean-Daniel Fekete 0.437107
Emmanuel Pietriga 0.260317
Yann Ponty 0.586364
Marc Schoenauer 0.443165
Franck Cappello 0.480488
Caroline Appert 0.273913
Michel Beaudouin-Lafon 0.297531
Wendy Mackay 0.302128
Anne Auger 0.591139
Evelyne Lutton 0.272973
Pierre Dragicevic 0.317284
Ioana Manolescu 0.507317
Nikolaus Hansen 0.686420
Nicolas Bredeche 0.544118
Olivier Teytaud 0.567925
François Goasdoué 0.430189
Nathalie Pernelle 0.482353
Fatiha Saïs 0.507317
Sarah Cohen-Boulakia 0.581818
Claude Marché 0.465957
Chantal Reynaud 0.513333
Olivier Chapuis 0.384615
Steven Martin 0.356410
Fatiha Zaidi 0.390625
Balázs Kégl 0.286842
Paola Tubaro 0.602564
Raymond Ros 0.655882
Cyril Furtlehner 0.666667
Anastasia Bezerianos 0.319403
Sylvie Boldo 0.488571
Guillaume Melquiond 0.472727
Marc Baboulin 0.540000
Dimo Brockhoff 0.610256
Nathann Cohen 0.380488
Petra Isenberg 0.479439
Tobias Isenberg 0.536752
Loïc Paulevé 0.726190
Céline Gicquel 0.818421
Isabelle Guyon 0.623596
Guillaume Charpiat 0.522581
Lonni Besançon 0.487879
Name: 0, dtype: float64
2_projection_nD scores:
None
umap matrices generated.
================================
Run params : {'3_2D__rkey': 'umap', '3_2D__rmetric': 'euclidean', '3_2D__rn_neighbors': 10, '3_2D__rmin_dist': 0.5, '3_2D__rinit': 'random', '3_2D__rlearning_rate': 1.0, '3_2D__rn_epochs': None, '3_2D__rrandom_state': None}
--------------------------------
Scoring params : {'3_2D__smin_score': 30, '3_2D__srecompute': True, '3_2D__ssample_size': None, '3_2D__sn_neighbors': 10, '3_2D__srandom_state': 42}
================================
----------- Scores -----------
3_2D__neighbors_articles_authors : 0.4078
3_2D__2_nD_3_2D_neighbors_articles_authors_dim_mean : -0.0701
3_2D__2_nD_3_2D_neighbors_articles_authors_dim_median : -0.0662
--------- Desc Scores ---------
3_2D__neighbors_articles_authors_det
0 Céline Gicquel : 0.63
1 Nikolaus Hansen : 0.61
2 Yann Ponty : 0.60
3 Sébastien Tixeuil : 0.59
4 Loïc Paulevé : 0.55
5 Dimo Brockhoff : 0.55
6 Cyril Furtlehner : 0.54
7 Nicolas Bredeche : 0.52
8 Raymond Ros : 0.52
9 Anne Auger : 0.52
10 Albert Cohen : 0.52
11 Philippe Caillou : 0.51
12 Paola Tubaro : 0.51
13 Olivier Teytaud : 0.51
14 Isabelle Guyon : 0.51
15 Guillaume Melquiond : 0.50
16 Chantal Reynaud : 0.47
17 Sarah Cohen-Boulakia : 0.45
18 Marc Baboulin : 0.45
19 Tobias Isenberg : 0.45
20 Sylvie Boldo : 0.45
21 Nathalie Pernelle : 0.45
22 Fatiha Saïs : 0.44
23 Nathann Cohen : 0.42
24 Franck Cappello : 0.41
25 Marc Schoenauer : 0.39
26 Lonni Besançon : 0.39
27 Ioana Manolescu : 0.39
28 Guillaume Charpiat : 0.39
29 Fatiha Zaidi : 0.37
30 Jean-Daniel Fekete : 0.37
31 Claude Marché : 0.36
32 Michèle Sebag : 0.35
33 Petra Isenberg : 0.33
34 Olivier Chapuis : 0.33
35 François Goasdoué : 0.31
36 Alain Denise : 0.31
37 Steven Martin : 0.30
38 Pierre Dragicevic : 0.27
39 Anastasia Bezerianos : 0.26
40 Michel Beaudouin-Lafon : 0.26
41 Wendy Mackay : 0.25
42 Johanne Cohen : 0.25
43 Caroline Appert : 0.23
44 Wendy E. Mackay : 0.22
45 Emmanuel Pietriga : 0.21
46 Evelyne Lutton : 0.20
47 Balázs Kégl : 0.19
dtype: object
VALUE : 0.4078
3_2D__2_nD_3_2D_neighbors_articles_authors_dim_det
0 Céline Gicquel : -0.19
1 Loïc Paulevé : -0.17
2 Philippe Caillou : -0.16
3 Petra Isenberg : -0.15
4 Guillaume Charpiat : -0.14
5 Raymond Ros : -0.13
6 Cyril Furtlehner : -0.13
7 Sarah Cohen-Boulakia : -0.13
8 François Goasdoué : -0.12
9 Ioana Manolescu : -0.12
10 Isabelle Guyon : -0.12
11 Claude Marché : -0.11
12 Lonni Besançon : -0.10
13 Balázs Kégl : -0.10
14 Sébastien Tixeuil : -0.09
15 Paola Tubaro : -0.09
16 Marc Baboulin : -0.09
17 Tobias Isenberg : -0.08
18 Nikolaus Hansen : -0.08
19 Evelyne Lutton : -0.08
20 Franck Cappello : -0.07
21 Jean-Daniel Fekete : -0.07
22 Fatiha Saïs : -0.07
23 Anne Auger : -0.07
24 Dimo Brockhoff : -0.06
25 Anastasia Bezerianos : -0.06
26 Steven Martin : -0.06
27 Olivier Teytaud : -0.06
28 Marc Schoenauer : -0.05
29 Olivier Chapuis : -0.05
30 Wendy Mackay : -0.05
31 Emmanuel Pietriga : -0.05
32 Chantal Reynaud : -0.04
33 Pierre Dragicevic : -0.04
34 Michel Beaudouin-Lafon : -0.04
35 Caroline Appert : -0.04
36 Michèle Sebag : -0.04
37 Sylvie Boldo : -0.04
38 Nathalie Pernelle : -0.04
39 Johanne Cohen : -0.03
40 Fatiha Zaidi : -0.02
41 Nicolas Bredeche : -0.02
42 Albert Cohen : -0.01
43 Alain Denise : 0.01
44 Yann Ponty : 0.01
45 Wendy E. Mackay : 0.02
46 Guillaume Melquiond : 0.03
47 Nathann Cohen : 0.04
dtype: object
VALUE : -0.0701
--------- Raw Scores ---------
3_2D__neighbors_articles_authors
Sébastien Tixeuil 0.587234
Michèle Sebag 0.345255
Johanne Cohen 0.246512
Albert Cohen 0.520690
Wendy E. Mackay 0.217391
Philippe Caillou 0.513953
Alain Denise 0.305556
Jean-Daniel Fekete 0.366038
Emmanuel Pietriga 0.212698
Yann Ponty 0.595455
Marc Schoenauer 0.389209
Franck Cappello 0.407317
Caroline Appert 0.234783
Michel Beaudouin-Lafon 0.256790
Wendy Mackay 0.248936
Anne Auger 0.522785
Evelyne Lutton 0.197297
Pierre Dragicevic 0.274074
Ioana Manolescu 0.387805
Nikolaus Hansen 0.607407
Nicolas Bredeche 0.523529
Olivier Teytaud 0.510377
François Goasdoué 0.305660
Nathalie Pernelle 0.447059
Fatiha Saïs 0.436585
Sarah Cohen-Boulakia 0.454545
Claude Marché 0.359574
Chantal Reynaud 0.470000
Olivier Chapuis 0.330769
Steven Martin 0.297436
Fatiha Zaidi 0.368750
Balázs Kégl 0.189474
Paola Tubaro 0.512821
Raymond Ros 0.523529
Cyril Furtlehner 0.535897
Anastasia Bezerianos 0.258209
Sylvie Boldo 0.451429
Guillaume Melquiond 0.500000
Marc Baboulin 0.453333
Dimo Brockhoff 0.546154
Nathann Cohen 0.417073
Petra Isenberg 0.330841
Tobias Isenberg 0.452137
Loïc Paulevé 0.554762
Céline Gicquel 0.626316
Isabelle Guyon 0.505618
Guillaume Charpiat 0.387097
Lonni Besançon 0.387879
dtype: float64
3_2D__2_nD_3_2D_neighbors_articles_authors_dim
2_nD__neighbors_articles_authors \
Sébastien Tixeuil 0.680851
Michèle Sebag 0.383942
Johanne Cohen 0.272093
Albert Cohen 0.529310
Wendy E. Mackay 0.200000
Philippe Caillou 0.674419
Alain Denise 0.297222
Jean-Daniel Fekete 0.437107
Emmanuel Pietriga 0.260317
Yann Ponty 0.586364
Marc Schoenauer 0.443165
Franck Cappello 0.480488
Caroline Appert 0.273913
Michel Beaudouin-Lafon 0.297531
Wendy Mackay 0.302128
Anne Auger 0.591139
Evelyne Lutton 0.272973
Pierre Dragicevic 0.317284
Ioana Manolescu 0.507317
Nikolaus Hansen 0.686420
Nicolas Bredeche 0.544118
Olivier Teytaud 0.567925
François Goasdoué 0.430189
Nathalie Pernelle 0.482353
Fatiha Saïs 0.507317
Sarah Cohen-Boulakia 0.581818
Claude Marché 0.465957
Chantal Reynaud 0.513333
Olivier Chapuis 0.384615
Steven Martin 0.356410
Fatiha Zaidi 0.390625
Balázs Kégl 0.286842
Paola Tubaro 0.602564
Raymond Ros 0.655882
Cyril Furtlehner 0.666667
Anastasia Bezerianos 0.319403
Sylvie Boldo 0.488571
Guillaume Melquiond 0.472727
Marc Baboulin 0.540000
Dimo Brockhoff 0.610256
Nathann Cohen 0.380488
Petra Isenberg 0.479439
Tobias Isenberg 0.536752
Loïc Paulevé 0.726190
Céline Gicquel 0.818421
Isabelle Guyon 0.623596
Guillaume Charpiat 0.522581
Lonni Besançon 0.487879
3_2D__neighbors_articles_authors scdec
Sébastien Tixeuil 0.587234 -0.093617
Michèle Sebag 0.345255 -0.038686
Johanne Cohen 0.246512 -0.025581
Albert Cohen 0.520690 -0.008621
Wendy E. Mackay 0.217391 0.017391
Philippe Caillou 0.513953 -0.160465
Alain Denise 0.305556 0.008333
Jean-Daniel Fekete 0.366038 -0.071069
Emmanuel Pietriga 0.212698 -0.047619
Yann Ponty 0.595455 0.009091
Marc Schoenauer 0.389209 -0.053957
Franck Cappello 0.407317 -0.073171
Caroline Appert 0.234783 -0.039130
Michel Beaudouin-Lafon 0.256790 -0.040741
Wendy Mackay 0.248936 -0.053191
Anne Auger 0.522785 -0.068354
Evelyne Lutton 0.197297 -0.075676
Pierre Dragicevic 0.274074 -0.043210
Ioana Manolescu 0.387805 -0.119512
Nikolaus Hansen 0.607407 -0.079012
Nicolas Bredeche 0.523529 -0.020588
Olivier Teytaud 0.510377 -0.057547
François Goasdoué 0.305660 -0.124528
Nathalie Pernelle 0.447059 -0.035294
Fatiha Saïs 0.436585 -0.070732
Sarah Cohen-Boulakia 0.454545 -0.127273
Claude Marché 0.359574 -0.106383
Chantal Reynaud 0.470000 -0.043333
Olivier Chapuis 0.330769 -0.053846
Steven Martin 0.297436 -0.058974
Fatiha Zaidi 0.368750 -0.021875
Balázs Kégl 0.189474 -0.097368
Paola Tubaro 0.512821 -0.089744
Raymond Ros 0.523529 -0.132353
Cyril Furtlehner 0.535897 -0.130769
Anastasia Bezerianos 0.258209 -0.061194
Sylvie Boldo 0.451429 -0.037143
Guillaume Melquiond 0.500000 0.027273
Marc Baboulin 0.453333 -0.086667
Dimo Brockhoff 0.546154 -0.064103
Nathann Cohen 0.417073 0.036585
Petra Isenberg 0.330841 -0.148598
Tobias Isenberg 0.452137 -0.084615
Loïc Paulevé 0.554762 -0.171429
Céline Gicquel 0.626316 -0.192105
Isabelle Guyon 0.505618 -0.117978
Guillaume Charpiat 0.387097 -0.135484
Lonni Besançon 0.387879 -0.100000
3_projection_2D scores:
None
Nothing in cache, initial Fitting with min_cluster_size=15 Found 53 clusters in 0.28661163699871395s
Max Fitting with min_cluster_size=30 Found 35 clusters in 0.10271434999958728s
Max Fitting with min_cluster_size=60 Found 4 clusters in 0.10759418200177606s
Midpoint Fitting with min_cluster_size=45 Found 17 clusters in 0.10179101500034449s
Midpoint Fitting with min_cluster_size=52 Found 4 clusters in 0.10810460799984867s
Re-Fitting with min_cluster_size=45 Found 17 clusters in 0.10246682100114413s
Clusters cached: [4, 4, 17, 35, 53]
Nothing in cache, initial Fitting with min_cluster_size=15 Found 53 clusters in 0.10724538099748315s
Max Fitting with min_cluster_size=30 Found 35 clusters in 0.10228354599894374s
Max Fitting with min_cluster_size=60 Found 4 clusters in 0.10733317900303518s
Midpoint Fitting with min_cluster_size=45 Found 17 clusters in 0.10190187199987122s
Midpoint Fitting with min_cluster_size=37 Found 27 clusters in 0.1012505369981227s
No need Re-Fitting with min_cluster_size=37
Clusters cached: [4, 17, 27, 35, 53]
================================
Run params : {'4_clus__rkey': 'hdbscan', '4_clus__rbase_factor': 3}
--------------------------------
Scoring params : {'4_clus__siter_stab': 2, '4_clus__sremove_stab': [0, 0.01, 0.03, 0.1, 0.25], '4_clus__smetric': 'euclidean', '4_clus__srandom_state': None}
================================
----------- Scores -----------
4_clus__nb_clust_0 : 8.0000
4_clus__silhouette_0 : -0.0402
4_clus__avg_word_couv_0 : 0.7029
4_clus__med_word_couv_0 : 0.7241
4_clus__avg_word_couv_minus_0 : 0.6734
4_clus__big_small_ratio_0 : 35.2500
4_clus__stab_clus_0 : 0.0059
4_clus__nb_clust_1 : 24.0000
4_clus__silhouette_1 : -0.1014
4_clus__avg_word_couv_1 : 0.6133
4_clus__med_word_couv_1 : 0.6122
4_clus__avg_word_couv_minus_1 : 0.5932
4_clus__big_small_ratio_1 : 54.0000
4_clus__stab_clus_1 : 0.0074
4_clus__avg_stab_avg : 0.0066
4_clus__avg_couv_avg : 0.6581
4_clus__clu_score : 0.3324
--------- Desc Scores ---------
4_clus__clus_eval_pos_0_det
0 networking internet architecture, ontology : s...
1 visualizations, interfaces : s 607 stb 0.10 + ...
2 biological, metabolic : s 147 stb 0.00 + 0.45 ...
3 compiler, optimizations : s 98 stb 0.00 + 0.65...
4 logic science, verification : s 97 stb 0.00 + ...
5 fluid, numerical simulations : s 95 stb 0.00 +...
6 vertex, minimum : s 94 stb 0.00 + 0.66 - 0.03
7 secondary structure, sequence : s 90 stb 0.00 ...
8 query : s 82 stb 0.00 + 0.99 - 0.03
9 testing, challenge : s 81 stb 0.00 + 0.80 - 0.06
10 monte carlo, games : s 77 stb 0.00 + 0.95 - 0.02
11 multi armed, adaptive : s 76 stb 0.00 + 0.84 -...
12 semantics, neural networks : s 62 stb 0.00 + 0...
13 trees, combinatorics : s 62 stb 0.00 + 0.71 - ...
14 multi objective, gradient : s 58 stb 0.00 + 0....
15 integer, lower bounds : s 50 stb 0.00 + 0.64 -...
16 floating point : s 48 stb 0.00 + 0.85 - 0.00
dtype: object
VALUE : 0.7029
4_clus__clus_eval_pos_1_det
0 humanities social sciences, ontologies : s 409...
1 systems biology, metabolism : s 147 stb 0.00 +...
2 information visualization, cognitive science :...
3 architectures, polyhedral : s 98 stb 0.00 + 0....
4 deductive, safety : s 97 stb 0.00 + 0.68 - 0.02
5 mechanics, flows : s 95 stb 0.00 + 0.57 - 0.02
6 vertices, computational complexity : s 94 stb ...
7 tangible, creative : s 91 stb 0.00 + 0.49 - 0.01
8 secondary, sequences : s 90 stb 0.00 + 0.69 - ...
9 internet architecture, simulation results : s ...
10 queries, query answers : s 82 stb 0.00 + 0.72 ...
11 conformance, automated machine learning : s 81...
12 monte carlo search, players : s 77 stb 0.00 + ...
13 multi armed bandit, multi armed bandits : s 76...
14 cloud, services : s 70 stb 0.00 + 0.69 - 0.04
15 calculus, object oriented : s 62 stb 0.00 + 0....
16 lattice, binary trees : s 62 stb 0.00 + 0.47 -...
17 multi objective optimization, stochastic gradi...
18 movement, spatial : s 52 stb 0.10 + 0.60 - 0.04
19 linear programming, bounds : s 50 stb 0.00 + 0...
20 floating point arithmetic, floating : s 48 stb...
21 visual analytics : s 44 stb 0.00 + 0.86 - 0.01
22 benchmarking : s 43 stb 0.00 + 0.86 - 0.01
23 energy, materialized views : s 41 stb 0.10 + 0...
24 scientific, provenance : s 39 stb 0.00 + 0.92 ...
25 image : s 39 stb 0.00 + 0.92 - 0.02
26 protein : s 37 stb 0.00 + 0.89 - 0.01
dtype: object
VALUE : 0.6133
--------- Raw Scores ---------
4_clustering scores:
None
['2_nD__neighbors_articles_authors', '3_2D__neighbors_articles_authors', '4_clus__clu_score']
================================
Run params : {}
--------------------------------
Scoring params : {'6_pst__sname_list': ['2_nD__neighbors_articles_authors', '3_2D__neighbors_articles_authors', '4_clus__clu_score']}
================================
----------- Scores -----------
6_pst__final_score : 0.4060
--------- Desc Scores ---------
6_pst__final_score_det
0 2_nD__neighbors_articles_authors : 0.48
1 3_2D__neighbors_articles_authors : 0.41
2 4_clus__clu_score : 0.33
dtype: object
VALUE : 0.4060
--------- Raw Scores ---------
6_post scores:
None
Finished running step!
.. GENERATED FROM PYTHON SOURCE LINES 406-407
When the experiment is run, results of all runs is saved in `experiment.results`. We access the values corresponding to each run with `experiment.results.runs_`.
.. GENERATED FROM PYTHON SOURCE LINES 407-425
.. code-block:: Python
experiment.results.runs_[0].scores
""
list(experiment.results.runs_[0].desc_scores.keys())
""
experiment.results.runs_[0].desc_scores
""
list(experiment.results.runs_[0].raw_scores.keys())
""
experiment.results.runs_[0].raw_scores
""
results.print_best(n=20)
.. rst-class:: sphx-glr-script-out
.. code-block:: none
------------------------
Best 20 results:
------------------------
0_agscore id 2_nD__rkey 2_nD__rnum_dims \
0 0.426323 68d289ffd85ae917e710 lsa 50
2 0.406000 a588bce09630b37c4111 lsa 50
1 0.405798 ce1c0d334fc75a6570a5 lsa 50
2_nD__rnormalize 2_nD__rextra_param 3_2D__rkey 3_2D__rmetric \
0 True True umap euclidean
2 True True umap euclidean
1 True True umap euclidean
3_2D__rn_neighbors 3_2D__rmin_dist 3_2D__rinit 3_2D__rlearning_rate \
0 10 0.10 random 1.0
2 10 0.50 random 1.0
1 10 0.25 random 1.0
3_2D__rn_epochs 3_2D__rrandom_state 4_clus__rkey 4_clus__rbase_factor \
0 None None hdbscan 3
2 None None hdbscan 3
1 None None hdbscan 3
2_nD__smin_score 2_nD__srecompute 2_nD__ssample_size 2_nD__sn_neighbors \
0 30 True None 10
2 30 True None 10
1 30 True None 10
2_nD__srandom_state 2_nD__sfactor 3_2D__smin_score 3_2D__srecompute \
0 42 10 30 True
2 42 10 30 True
1 42 10 30 True
3_2D__ssample_size 3_2D__sn_neighbors 3_2D__srandom_state \
0 None 10 42
2 None 10 42
1 None 10 42
4_clus__siter_stab 4_clus__sremove_stab 4_clus__smetric \
0 2 [0, 0.01, 0.03, 0.1, 0.25] euclidean
2 2 [0, 0.01, 0.03, 0.1, 0.25] euclidean
1 2 [0, 0.01, 0.03, 0.1, 0.25] euclidean
4_clus__srandom_state 6_pst__sname_list \
0 None [2_nD__neighbors_articles_authors, 3_2D__neigh...
2 None [2_nD__neighbors_articles_authors, 3_2D__neigh...
1 None [2_nD__neighbors_articles_authors, 3_2D__neigh...
2_nD__neighbors_articles_authors 2_nD__custom_score_lsa \
0 0.477852 10
2 0.477852 10
1 0.477852 10
3_2D__neighbors_articles_authors \
0 0.432393
2 0.407792
1 0.425460
3_2D__2_nD_3_2D_neighbors_articles_authors_dim_mean \
0 -0.045459
2 -0.070060
1 -0.052393
3_2D__2_nD_3_2D_neighbors_articles_authors_dim_median 4_clus__nb_clust_0 \
0 -0.044640 8
2 -0.066228 8
1 -0.046651 8
4_clus__silhouette_0 4_clus__avg_word_couv_0 4_clus__med_word_couv_0 \
0 0.160530 0.443784 0.465649
2 -0.040228 0.702874 0.724138
1 -0.054422 0.517312 0.537415
4_clus__avg_word_couv_minus_0 4_clus__big_small_ratio_0 \
0 0.414292 6.209424
2 0.673413 35.250000
1 0.488406 14.324561
4_clus__stab_clus_0 4_clus__nb_clust_1 4_clus__silhouette_1 \
0 0.233333 24 0.247894
2 0.005882 24 -0.101386
1 0.044444 24 0.056935
4_clus__avg_word_couv_1 4_clus__med_word_couv_1 \
0 0.625363 0.604839
2 0.613262 0.612245
1 0.646574 0.696078
4_clus__avg_word_couv_minus_1 4_clus__big_small_ratio_1 \
0 0.602092 16.894737
2 0.593190 54.000000
1 0.620567 26.283019
4_clus__stab_clus_1 4_clus__avg_stab_avg 4_clus__avg_couv_avg \
0 0.172414 0.202874 0.534574
2 0.007407 0.006645 0.658068
1 0.048000 0.046222 0.581943
4_clus__clu_score 6_pst__final_score \
0 0.368724 0.426323
2 0.332356 0.406000
1 0.314082 0.405798
dump active
0 experiment_custom/lisn/2022.11.15.1/0/mat_arti... True
2 experiment_custom/lisn/2022.11.15.1/0/mat_arti... True
1 experiment_custom/lisn/2022.11.15.1/0/mat_arti... True
.. GENERATED FROM PYTHON SOURCE LINES 426-429
Let's see some of the results of the experiment from the file system.
We will first check the contents of the `scores` directory.
.. GENERATED FROM PYTHON SOURCE LINES 429-432
.. code-block:: Python
# !ls $TOP_DIR/scores
.. GENERATED FROM PYTHON SOURCE LINES 433-434
`6_pst__final_scores.csv` file contains the final scores for each set of parameters together with the parameter values. We had 3 runs, so there are 3 values.
.. GENERATED FROM PYTHON SOURCE LINES 434-437
.. code-block:: Python
# !cat $TOP_DIR/scores/6_pst__final_score.csv
.. GENERATED FROM PYTHON SOURCE LINES 438-439
The `final_results.csv` displays each score calculated during the experiment in separate columns together with `rank` and an aggregated score `agscore`. This value is the same as the value in `6_post__final_score.csv` file.
.. GENERATED FROM PYTHON SOURCE LINES 439-442
.. code-block:: Python
# !cat $TOP_DIR/scores/final_results.csv
.. GENERATED FROM PYTHON SOURCE LINES 443-444
The other files in the directory contains the single score calculated for all runs. For example `2_nD__neighbors_articles_authors.csv` file contains the `2_nD__neighbors_articles_authors` scores for 3 runs.
.. GENERATED FROM PYTHON SOURCE LINES 444-447
.. code-block:: Python
# !cat $TOP_DIR/scores/2_nD__neighbors_articles_authors.csv
.. GENERATED FROM PYTHON SOURCE LINES 448-449
The files that contain `det` in theirs names, contains the neighbors and their scores used to calculate the score `2_nD__neighbors_articles_authors` for each run.
.. GENERATED FROM PYTHON SOURCE LINES 449-452
.. code-block:: Python
# !cat $TOP_DIR/scores/2_nD__neighbors_articles_authors_det.csv
.. GENERATED FROM PYTHON SOURCE LINES 453-456
These files also reside in the hierarchical dataset directories generated during the run.
For example `experiment_custom/lisn/2022.11.15.1/0/mat_articles__authors_4_teams_4_labs_4_words_10_0.05_None_None_5_4/lsa_50_True_True/scores_2_nD__neighbors_articles_authors_det.csv` file, but only for the specific run together with hyperparameters and scoring parameters.
.. GENERATED FROM PYTHON SOURCE LINES 456-459
.. code-block:: Python
# !cat $TOP_DIR/lisn/2022.11.15.1/0/mat_articles__authors_4_teams_4_labs_4_words_10_0.05_None_None_5_4/lsa_50_True_True/scores_2_nD__neighbors_articles_authors_det.csv
.. GENERATED FROM PYTHON SOURCE LINES 460-465
Actually for each set of parameters, the estimator generates a directory structure of the form:
`top_dir / dataset / dataset_version / robustseed / dataset_column_parameters / projection_nd_key_dim / projection2D_key_n_neighbors_min_dist_metric_init_learning_rate_repulsion_strength / clustering_key_base_factor`.
Each score calculated at a certain level in the directory structure is saved in that directory.
.. GENERATED FROM PYTHON SOURCE LINES 467-468
Now, we will continue the experiment to run for 3 more set of parameters.
.. GENERATED FROM PYTHON SOURCE LINES 468-471
.. code-block:: Python
experiment.run(3)
.. rst-class:: sphx-glr-script-out
.. code-block:: none
Running experiment for parameters:
{'id': '06128c38fd13123e263c', 'robustseed': 0, 'authors__filter_min_score': 4, 'filter_min_score': 6, 'projection_nD': {'key': 'lsa', 'num_dims': 50, 'extra_param': True}, 'projection_2D': {'key': 'umap', 'n_neighbors': 20, 'min_dist': 0.1, 'metric': 'euclidean'}}
Matrices of type mat for articles, authors, teams, labs, words already exist. Skipping creation.
Matrices of type lsa for articles, authors, teams, labs, words already exist. Skipping creation.
================================
Run params : {'2_nD__rkey': 'lsa', '2_nD__rnum_dims': 50, '2_nD__rnormalize': True, '2_nD__rextra_param': True}
--------------------------------
Scoring params : {'2_nD__smin_score': 30, '2_nD__srecompute': True, '2_nD__ssample_size': None, '2_nD__sn_neighbors': 10, '2_nD__srandom_state': 42}
================================
----------- Scores -----------
2_nD__neighbors_articles_authors : 0.4779
--------- Desc Scores ---------
2_nD__neighbors_articles_authors_det
0 Céline Gicquel : 0.82
1 Loïc Paulevé : 0.73
2 Nikolaus Hansen : 0.69
3 Sébastien Tixeuil : 0.68
4 Philippe Caillou : 0.67
5 Cyril Furtlehner : 0.67
6 Raymond Ros : 0.66
7 Isabelle Guyon : 0.62
8 Dimo Brockhoff : 0.61
9 Paola Tubaro : 0.60
10 Anne Auger : 0.59
11 Yann Ponty : 0.59
12 Sarah Cohen-Boulakia : 0.58
13 Olivier Teytaud : 0.57
14 Nicolas Bredeche : 0.54
15 Marc Baboulin : 0.54
16 Tobias Isenberg : 0.54
17 Albert Cohen : 0.53
18 Guillaume Charpiat : 0.52
19 Chantal Reynaud : 0.51
20 Fatiha Saïs : 0.51
21 Ioana Manolescu : 0.51
22 Sylvie Boldo : 0.49
23 Lonni Besançon : 0.49
24 Nathalie Pernelle : 0.48
25 Franck Cappello : 0.48
26 Petra Isenberg : 0.48
27 Guillaume Melquiond : 0.47
28 Claude Marché : 0.47
29 Marc Schoenauer : 0.44
30 Jean-Daniel Fekete : 0.44
31 François Goasdoué : 0.43
32 Fatiha Zaidi : 0.39
33 Olivier Chapuis : 0.38
34 Michèle Sebag : 0.38
35 Nathann Cohen : 0.38
36 Steven Martin : 0.36
37 Anastasia Bezerianos : 0.32
38 Pierre Dragicevic : 0.32
39 Wendy Mackay : 0.30
40 Michel Beaudouin-Lafon : 0.30
41 Alain Denise : 0.30
42 Balázs Kégl : 0.29
43 Caroline Appert : 0.27
44 Evelyne Lutton : 0.27
45 Johanne Cohen : 0.27
46 Emmanuel Pietriga : 0.26
47 Wendy E. Mackay : 0.20
Name: 0, dtype: object
VALUE : 0.4779
--------- Raw Scores ---------
2_nD__neighbors_articles_authors
Sébastien Tixeuil 0.680851
Michèle Sebag 0.383942
Johanne Cohen 0.272093
Albert Cohen 0.529310
Wendy E. Mackay 0.200000
Philippe Caillou 0.674419
Alain Denise 0.297222
Jean-Daniel Fekete 0.437107
Emmanuel Pietriga 0.260317
Yann Ponty 0.586364
Marc Schoenauer 0.443165
Franck Cappello 0.480488
Caroline Appert 0.273913
Michel Beaudouin-Lafon 0.297531
Wendy Mackay 0.302128
Anne Auger 0.591139
Evelyne Lutton 0.272973
Pierre Dragicevic 0.317284
Ioana Manolescu 0.507317
Nikolaus Hansen 0.686420
Nicolas Bredeche 0.544118
Olivier Teytaud 0.567925
François Goasdoué 0.430189
Nathalie Pernelle 0.482353
Fatiha Saïs 0.507317
Sarah Cohen-Boulakia 0.581818
Claude Marché 0.465957
Chantal Reynaud 0.513333
Olivier Chapuis 0.384615
Steven Martin 0.356410
Fatiha Zaidi 0.390625
Balázs Kégl 0.286842
Paola Tubaro 0.602564
Raymond Ros 0.655882
Cyril Furtlehner 0.666667
Anastasia Bezerianos 0.319403
Sylvie Boldo 0.488571
Guillaume Melquiond 0.472727
Marc Baboulin 0.540000
Dimo Brockhoff 0.610256
Nathann Cohen 0.380488
Petra Isenberg 0.479439
Tobias Isenberg 0.536752
Loïc Paulevé 0.726190
Céline Gicquel 0.818421
Isabelle Guyon 0.623596
Guillaume Charpiat 0.522581
Lonni Besançon 0.487879
Name: 0, dtype: float64
================================
Run params : {}
--------------------------------
Scoring params : {}
================================
----------- Scores -----------
custom_score_lsa : 10.0000
--------- Desc Scores ---------
--------- Raw Scores ---------
custom scores:
None
================================
Run params : {'2_nD__rkey': 'lsa', '2_nD__rnum_dims': 50, '2_nD__rnormalize': True, '2_nD__rextra_param': True}
--------------------------------
Scoring params : {'2_nD__smin_score': 30, '2_nD__srecompute': True, '2_nD__ssample_size': None, '2_nD__sn_neighbors': 10, '2_nD__srandom_state': 42, '2_nD__sfactor': 10}
================================
----------- Scores -----------
2_nD__neighbors_articles_authors : 0.4779
2_nD__custom_score_lsa : 10.0000
--------- Desc Scores ---------
2_nD__neighbors_articles_authors_det
0 Céline Gicquel : 0.82
1 Loïc Paulevé : 0.73
2 Nikolaus Hansen : 0.69
3 Sébastien Tixeuil : 0.68
4 Philippe Caillou : 0.67
5 Cyril Furtlehner : 0.67
6 Raymond Ros : 0.66
7 Isabelle Guyon : 0.62
8 Dimo Brockhoff : 0.61
9 Paola Tubaro : 0.60
10 Anne Auger : 0.59
11 Yann Ponty : 0.59
12 Sarah Cohen-Boulakia : 0.58
13 Olivier Teytaud : 0.57
14 Nicolas Bredeche : 0.54
15 Marc Baboulin : 0.54
16 Tobias Isenberg : 0.54
17 Albert Cohen : 0.53
18 Guillaume Charpiat : 0.52
19 Chantal Reynaud : 0.51
20 Fatiha Saïs : 0.51
21 Ioana Manolescu : 0.51
22 Sylvie Boldo : 0.49
23 Lonni Besançon : 0.49
24 Nathalie Pernelle : 0.48
25 Franck Cappello : 0.48
26 Petra Isenberg : 0.48
27 Guillaume Melquiond : 0.47
28 Claude Marché : 0.47
29 Marc Schoenauer : 0.44
30 Jean-Daniel Fekete : 0.44
31 François Goasdoué : 0.43
32 Fatiha Zaidi : 0.39
33 Olivier Chapuis : 0.38
34 Michèle Sebag : 0.38
35 Nathann Cohen : 0.38
36 Steven Martin : 0.36
37 Anastasia Bezerianos : 0.32
38 Pierre Dragicevic : 0.32
39 Wendy Mackay : 0.30
40 Michel Beaudouin-Lafon : 0.30
41 Alain Denise : 0.30
42 Balázs Kégl : 0.29
43 Caroline Appert : 0.27
44 Evelyne Lutton : 0.27
45 Johanne Cohen : 0.27
46 Emmanuel Pietriga : 0.26
47 Wendy E. Mackay : 0.20
Name: 0, dtype: object
VALUE : 0.4779
--------- Raw Scores ---------
2_nD__neighbors_articles_authors
Sébastien Tixeuil 0.680851
Michèle Sebag 0.383942
Johanne Cohen 0.272093
Albert Cohen 0.529310
Wendy E. Mackay 0.200000
Philippe Caillou 0.674419
Alain Denise 0.297222
Jean-Daniel Fekete 0.437107
Emmanuel Pietriga 0.260317
Yann Ponty 0.586364
Marc Schoenauer 0.443165
Franck Cappello 0.480488
Caroline Appert 0.273913
Michel Beaudouin-Lafon 0.297531
Wendy Mackay 0.302128
Anne Auger 0.591139
Evelyne Lutton 0.272973
Pierre Dragicevic 0.317284
Ioana Manolescu 0.507317
Nikolaus Hansen 0.686420
Nicolas Bredeche 0.544118
Olivier Teytaud 0.567925
François Goasdoué 0.430189
Nathalie Pernelle 0.482353
Fatiha Saïs 0.507317
Sarah Cohen-Boulakia 0.581818
Claude Marché 0.465957
Chantal Reynaud 0.513333
Olivier Chapuis 0.384615
Steven Martin 0.356410
Fatiha Zaidi 0.390625
Balázs Kégl 0.286842
Paola Tubaro 0.602564
Raymond Ros 0.655882
Cyril Furtlehner 0.666667
Anastasia Bezerianos 0.319403
Sylvie Boldo 0.488571
Guillaume Melquiond 0.472727
Marc Baboulin 0.540000
Dimo Brockhoff 0.610256
Nathann Cohen 0.380488
Petra Isenberg 0.479439
Tobias Isenberg 0.536752
Loïc Paulevé 0.726190
Céline Gicquel 0.818421
Isabelle Guyon 0.623596
Guillaume Charpiat 0.522581
Lonni Besançon 0.487879
Name: 0, dtype: float64
2_projection_nD scores:
None
umap matrices generated.
================================
Run params : {'3_2D__rkey': 'umap', '3_2D__rmetric': 'euclidean', '3_2D__rn_neighbors': 20, '3_2D__rmin_dist': 0.1, '3_2D__rinit': 'random', '3_2D__rlearning_rate': 1.0, '3_2D__rn_epochs': None, '3_2D__rrandom_state': None}
--------------------------------
Scoring params : {'3_2D__smin_score': 30, '3_2D__srecompute': True, '3_2D__ssample_size': None, '3_2D__sn_neighbors': 10, '3_2D__srandom_state': 42}
================================
----------- Scores -----------
3_2D__neighbors_articles_authors : 0.4228
3_2D__2_nD_3_2D_neighbors_articles_authors_dim_mean : -0.0551
3_2D__2_nD_3_2D_neighbors_articles_authors_dim_median : -0.0494
--------- Desc Scores ---------
3_2D__neighbors_articles_authors_det
0 Loïc Paulevé : 0.71
1 Céline Gicquel : 0.65
2 Cyril Furtlehner : 0.61
3 Dimo Brockhoff : 0.60
4 Nikolaus Hansen : 0.60
5 Isabelle Guyon : 0.59
6 Philippe Caillou : 0.59
7 Sébastien Tixeuil : 0.58
8 Raymond Ros : 0.56
9 Yann Ponty : 0.55
10 Tobias Isenberg : 0.54
11 Olivier Teytaud : 0.54
12 Anne Auger : 0.51
13 Albert Cohen : 0.51
14 Marc Baboulin : 0.48
15 Ioana Manolescu : 0.47
16 Franck Cappello : 0.47
17 Nicolas Bredeche : 0.46
18 Fatiha Saïs : 0.45
19 Sylvie Boldo : 0.44
20 Nathalie Pernelle : 0.43
21 Guillaume Melquiond : 0.43
22 Paola Tubaro : 0.43
23 Chantal Reynaud : 0.42
24 Marc Schoenauer : 0.42
25 Jean-Daniel Fekete : 0.42
26 Nathann Cohen : 0.41
27 Claude Marché : 0.41
28 Fatiha Zaidi : 0.41
29 Sarah Cohen-Boulakia : 0.40
30 Lonni Besançon : 0.40
31 Petra Isenberg : 0.39
32 Michèle Sebag : 0.37
33 Guillaume Charpiat : 0.35
34 François Goasdoué : 0.34
35 Steven Martin : 0.34
36 Pierre Dragicevic : 0.32
37 Olivier Chapuis : 0.32
38 Michel Beaudouin-Lafon : 0.31
39 Alain Denise : 0.29
40 Anastasia Bezerianos : 0.29
41 Johanne Cohen : 0.22
42 Caroline Appert : 0.22
43 Wendy Mackay : 0.22
44 Emmanuel Pietriga : 0.22
45 Evelyne Lutton : 0.21
46 Wendy E. Mackay : 0.20
47 Balázs Kégl : 0.20
dtype: object
VALUE : 0.4228
3_2D__2_nD_3_2D_neighbors_articles_authors_dim_det
0 Sarah Cohen-Boulakia : -0.18
1 Paola Tubaro : -0.17
2 Guillaume Charpiat : -0.17
3 Céline Gicquel : -0.17
4 Sébastien Tixeuil : -0.10
5 Nikolaus Hansen : -0.09
6 Raymond Ros : -0.09
7 Petra Isenberg : -0.09
8 François Goasdoué : -0.09
9 Chantal Reynaud : -0.09
10 Balázs Kégl : -0.09
11 Philippe Caillou : -0.09
12 Lonni Besançon : -0.09
13 Anne Auger : -0.08
14 Wendy Mackay : -0.08
15 Nicolas Bredeche : -0.08
16 Evelyne Lutton : -0.06
17 Marc Baboulin : -0.06
18 Olivier Chapuis : -0.06
19 Cyril Furtlehner : -0.06
20 Claude Marché : -0.06
21 Fatiha Saïs : -0.05
22 Caroline Appert : -0.05
23 Nathalie Pernelle : -0.05
24 Johanne Cohen : -0.05
25 Sylvie Boldo : -0.05
26 Emmanuel Pietriga : -0.04
27 Guillaume Melquiond : -0.04
28 Ioana Manolescu : -0.04
29 Yann Ponty : -0.04
30 Isabelle Guyon : -0.03
31 Olivier Teytaud : -0.03
32 Anastasia Bezerianos : -0.03
33 Albert Cohen : -0.02
34 Steven Martin : -0.02
35 Marc Schoenauer : -0.02
36 Jean-Daniel Fekete : -0.02
37 Loïc Paulevé : -0.02
38 Franck Cappello : -0.01
39 Michèle Sebag : -0.01
40 Dimo Brockhoff : -0.01
41 Alain Denise : -0.00
42 Tobias Isenberg : 0.00
43 Wendy E. Mackay : 0.00
44 Pierre Dragicevic : 0.01
45 Michel Beaudouin-Lafon : 0.01
46 Fatiha Zaidi : 0.02
47 Nathann Cohen : 0.03
dtype: object
VALUE : -0.0551
--------- Raw Scores ---------
3_2D__neighbors_articles_authors
Sébastien Tixeuil 0.578723
Michèle Sebag 0.372263
Johanne Cohen 0.223256
Albert Cohen 0.505172
Wendy E. Mackay 0.204348
Philippe Caillou 0.586047
Alain Denise 0.294444
Jean-Daniel Fekete 0.417610
Emmanuel Pietriga 0.217460
Yann Ponty 0.546970
Marc Schoenauer 0.423022
Franck Cappello 0.465854
Caroline Appert 0.221739
Michel Beaudouin-Lafon 0.306173
Wendy Mackay 0.221277
Anne Auger 0.510127
Evelyne Lutton 0.208108
Pierre Dragicevic 0.324691
Ioana Manolescu 0.467073
Nikolaus Hansen 0.595062
Nicolas Bredeche 0.464706
Olivier Teytaud 0.535849
François Goasdoué 0.339623
Nathalie Pernelle 0.432353
Fatiha Saïs 0.453659
Sarah Cohen-Boulakia 0.403030
Claude Marché 0.408511
Chantal Reynaud 0.423333
Olivier Chapuis 0.323077
Steven Martin 0.335897
Fatiha Zaidi 0.406250
Balázs Kégl 0.197368
Paola Tubaro 0.428205
Raymond Ros 0.564706
Cyril Furtlehner 0.607692
Anastasia Bezerianos 0.292537
Sylvie Boldo 0.440000
Guillaume Melquiond 0.430303
Marc Baboulin 0.477778
Dimo Brockhoff 0.602564
Nathann Cohen 0.409756
Petra Isenberg 0.388785
Tobias Isenberg 0.536752
Loïc Paulevé 0.707143
Céline Gicquel 0.652632
Isabelle Guyon 0.588764
Guillaume Charpiat 0.351613
Lonni Besançon 0.400000
dtype: float64
3_2D__2_nD_3_2D_neighbors_articles_authors_dim
2_nD__neighbors_articles_authors \
Sébastien Tixeuil 0.680851
Michèle Sebag 0.383942
Johanne Cohen 0.272093
Albert Cohen 0.529310
Wendy E. Mackay 0.200000
Philippe Caillou 0.674419
Alain Denise 0.297222
Jean-Daniel Fekete 0.437107
Emmanuel Pietriga 0.260317
Yann Ponty 0.586364
Marc Schoenauer 0.443165
Franck Cappello 0.480488
Caroline Appert 0.273913
Michel Beaudouin-Lafon 0.297531
Wendy Mackay 0.302128
Anne Auger 0.591139
Evelyne Lutton 0.272973
Pierre Dragicevic 0.317284
Ioana Manolescu 0.507317
Nikolaus Hansen 0.686420
Nicolas Bredeche 0.544118
Olivier Teytaud 0.567925
François Goasdoué 0.430189
Nathalie Pernelle 0.482353
Fatiha Saïs 0.507317
Sarah Cohen-Boulakia 0.581818
Claude Marché 0.465957
Chantal Reynaud 0.513333
Olivier Chapuis 0.384615
Steven Martin 0.356410
Fatiha Zaidi 0.390625
Balázs Kégl 0.286842
Paola Tubaro 0.602564
Raymond Ros 0.655882
Cyril Furtlehner 0.666667
Anastasia Bezerianos 0.319403
Sylvie Boldo 0.488571
Guillaume Melquiond 0.472727
Marc Baboulin 0.540000
Dimo Brockhoff 0.610256
Nathann Cohen 0.380488
Petra Isenberg 0.479439
Tobias Isenberg 0.536752
Loïc Paulevé 0.726190
Céline Gicquel 0.818421
Isabelle Guyon 0.623596
Guillaume Charpiat 0.522581
Lonni Besançon 0.487879
3_2D__neighbors_articles_authors scdec
Sébastien Tixeuil 0.578723 -0.102128
Michèle Sebag 0.372263 -0.011679
Johanne Cohen 0.223256 -0.048837
Albert Cohen 0.505172 -0.024138
Wendy E. Mackay 0.204348 0.004348
Philippe Caillou 0.586047 -0.088372
Alain Denise 0.294444 -0.002778
Jean-Daniel Fekete 0.417610 -0.019497
Emmanuel Pietriga 0.217460 -0.042857
Yann Ponty 0.546970 -0.039394
Marc Schoenauer 0.423022 -0.020144
Franck Cappello 0.465854 -0.014634
Caroline Appert 0.221739 -0.052174
Michel Beaudouin-Lafon 0.306173 0.008642
Wendy Mackay 0.221277 -0.080851
Anne Auger 0.510127 -0.081013
Evelyne Lutton 0.208108 -0.064865
Pierre Dragicevic 0.324691 0.007407
Ioana Manolescu 0.467073 -0.040244
Nikolaus Hansen 0.595062 -0.091358
Nicolas Bredeche 0.464706 -0.079412
Olivier Teytaud 0.535849 -0.032075
François Goasdoué 0.339623 -0.090566
Nathalie Pernelle 0.432353 -0.050000
Fatiha Saïs 0.453659 -0.053659
Sarah Cohen-Boulakia 0.403030 -0.178788
Claude Marché 0.408511 -0.057447
Chantal Reynaud 0.423333 -0.090000
Olivier Chapuis 0.323077 -0.061538
Steven Martin 0.335897 -0.020513
Fatiha Zaidi 0.406250 0.015625
Balázs Kégl 0.197368 -0.089474
Paola Tubaro 0.428205 -0.174359
Raymond Ros 0.564706 -0.091176
Cyril Furtlehner 0.607692 -0.058974
Anastasia Bezerianos 0.292537 -0.026866
Sylvie Boldo 0.440000 -0.048571
Guillaume Melquiond 0.430303 -0.042424
Marc Baboulin 0.477778 -0.062222
Dimo Brockhoff 0.602564 -0.007692
Nathann Cohen 0.409756 0.029268
Petra Isenberg 0.388785 -0.090654
Tobias Isenberg 0.536752 0.000000
Loïc Paulevé 0.707143 -0.019048
Céline Gicquel 0.652632 -0.165789
Isabelle Guyon 0.588764 -0.034831
Guillaume Charpiat 0.351613 -0.170968
Lonni Besançon 0.400000 -0.087879
3_projection_2D scores:
None
Nothing in cache, initial Fitting with min_cluster_size=15 Found 78 clusters in 0.2512342800000624s
Max Fitting with min_cluster_size=30 Found 52 clusters in 0.1034947119987919s
Max Fitting with min_cluster_size=60 Found 23 clusters in 0.1068140620009217s
Max Fitting with min_cluster_size=120 Found 3 clusters in 0.10146449500098242s
Midpoint Fitting with min_cluster_size=90 Found 5 clusters in 0.10416944600001443s
Midpoint Fitting with min_cluster_size=75 Found 18 clusters in 0.09905218700077967s
Midpoint Fitting with min_cluster_size=82 Found 5 clusters in 0.09654392100128462s
Re-Fitting with min_cluster_size=75 Found 18 clusters in 0.09632515700286604s
Clusters cached: [3, 5, 5, 18, 23, 52, 78]
Nothing in cache, initial Fitting with min_cluster_size=15 Found 78 clusters in 0.10544126700187917s
Max Fitting with min_cluster_size=30 Found 52 clusters in 0.10392791900085285s
Max Fitting with min_cluster_size=60 Found 23 clusters in 0.10288754399880418s
Midpoint Fitting with min_cluster_size=45 Found 32 clusters in 0.09387680500003626s
Midpoint Fitting with min_cluster_size=52 Found 30 clusters in 0.10210112799904891s
No need Re-Fitting with min_cluster_size=52
Clusters cached: [23, 30, 32, 52, 78]
================================
Run params : {'4_clus__rkey': 'hdbscan', '4_clus__rbase_factor': 3}
--------------------------------
Scoring params : {'4_clus__siter_stab': 2, '4_clus__sremove_stab': [0, 0.01, 0.03, 0.1, 0.25], '4_clus__smetric': 'euclidean', '4_clus__srandom_state': None}
================================
----------- Scores -----------
4_clus__nb_clust_0 : 8.0000
4_clus__silhouette_0 : 0.2633
4_clus__avg_word_couv_0 : 0.6128
4_clus__med_word_couv_0 : 0.5791
4_clus__avg_word_couv_minus_0 : 0.5838
4_clus__big_small_ratio_0 : 9.9200
4_clus__stab_clus_0 : 0.0667
4_clus__nb_clust_1 : 24.0000
4_clus__silhouette_1 : 0.3176
4_clus__avg_word_couv_1 : 0.5868
4_clus__med_word_couv_1 : 0.5507
4_clus__avg_word_couv_minus_1 : 0.5636
4_clus__big_small_ratio_1 : 13.7170
4_clus__stab_clus_1 : 0.0333
4_clus__avg_stab_avg : 0.0500
4_clus__avg_couv_avg : 0.5998
4_clus__clu_score : 0.3249
--------- Desc Scores ---------
4_clus__clus_eval_pos_0_det
0 visualizations, interfaces : s 535 stb 0.60 + ...
1 stochastic, neural networks : s 501 stb 0.00 +...
2 query, ontology : s 347 stb 0.20 + 0.57 - 0.03
3 logic science, semantics : s 225 stb 0.00 + 0....
4 visual analytics, humanities social sciences :...
5 networking internet architecture, services : s...
6 architectures, programming languages : s 192 s...
7 biology, biological : s 191 stb 0.00 + 0.53 - ...
8 french, computation language : s 138 stb 0.00 ...
9 fluid, numerical simulations : s 130 stb 0.00 ...
10 vertices, minimum : s 129 stb 0.00 + 0.68 - 0.03
11 combinatorics, automata : s 126 stb 0.00 + 0.4...
12 image, motion : s 113 stb 0.00 + 0.77 - 0.03
13 floating point, benchmarking : s 108 stb 0.30 ...
14 secondary structure, sequence : s 99 stb 0.00 ...
15 cluster computing, cloud : s 97 stb 0.00 + 0.7...
16 agent, scientific : s 92 stb 0.00 + 0.85 - 0.05
17 monte carlo, games : s 75 stb 0.00 + 0.96 - 0.02
dtype: object
VALUE : 0.6128
4_clus__clus_eval_pos_1_det
0 queries, ontologies : s 347 stb 0.00 + 0.48 - ...
1 verification, proof assistant : s 225 stb 0.00...
2 internet architecture, service : s 209 stb 0.0...
3 molecular, genes : s 191 stb 0.00 + 0.46 - 0.01
4 natural language, translation : s 138 stb 0.00...
5 display, pointing : s 137 stb 0.00 + 0.48 - 0.02
6 dynamics, mechanics : s 130 stb 0.00 + 0.62 - ...
7 social sciences, social networks : s 130 stb 0...
8 compiler, parallelism : s 130 stb 0.00 + 0.56 ...
9 inference, traffic : s 129 stb 0.00 + 0.50 - 0.03
10 discrete mathematics, vertex : s 129 stb 0.00 ...
11 finite, trees : s 126 stb 0.10 + 0.52 - 0.06
12 information visualization, cognitive science :...
13 image processing, vision pattern recognition :...
14 floating point arithmetic, black optimization ...
15 multi objective, neural network : s 104 stb 0....
16 secondary, sampling : s 99 stb 0.00 + 0.72 - 0.03
17 fault, cloud computing : s 97 stb 0.00 + 0.64 ...
18 multi agent, scientific workflows : s 92 stb 0...
19 creative, musical : s 82 stb 0.00 + 0.48 - 0.01
20 optimization control, neural evolutionary comp...
21 virtual, augmented reality : s 77 stb 0.00 + 0...
22 monte carlo search, players : s 75 stb 0.00 + ...
23 testing, mining : s 74 stb 0.40 + 0.89 - 0.03
24 analytics : s 73 stb 0.00 + 0.84 - 0.01
25 stabilizing, population protocols : s 71 stb 0...
26 multi armed, discrete event systems : s 69 stb...
27 operations, mixed integer linear programming :...
28 linear systems, matrices : s 62 stb 0.00 + 0.7...
29 challenge, classification : s 53 stb 0.00 + 0....
dtype: object
VALUE : 0.5868
--------- Raw Scores ---------
4_clustering scores:
None
['2_nD__neighbors_articles_authors', '3_2D__neighbors_articles_authors', '4_clus__clu_score']
================================
Run params : {}
--------------------------------
Scoring params : {'6_pst__sname_list': ['2_nD__neighbors_articles_authors', '3_2D__neighbors_articles_authors', '4_clus__clu_score']}
================================
----------- Scores -----------
6_pst__final_score : 0.4085
--------- Desc Scores ---------
6_pst__final_score_det
0 2_nD__neighbors_articles_authors : 0.48
1 3_2D__neighbors_articles_authors : 0.42
2 4_clus__clu_score : 0.32
dtype: object
VALUE : 0.4085
--------- Raw Scores ---------
6_post scores:
None
Finished running step!
Running experiment for parameters:
{'id': '1040be9e44c8ab04aa0e', 'robustseed': 0, 'authors__filter_min_score': 4, 'filter_min_score': 6, 'projection_nD': {'key': 'lsa', 'num_dims': 50, 'extra_param': True}, 'projection_2D': {'key': 'umap', 'n_neighbors': 20, 'min_dist': 0.25, 'metric': 'euclidean'}}
Matrices of type mat for articles, authors, teams, labs, words already exist. Skipping creation.
Matrices of type lsa for articles, authors, teams, labs, words already exist. Skipping creation.
================================
Run params : {'2_nD__rkey': 'lsa', '2_nD__rnum_dims': 50, '2_nD__rnormalize': True, '2_nD__rextra_param': True}
--------------------------------
Scoring params : {'2_nD__smin_score': 30, '2_nD__srecompute': True, '2_nD__ssample_size': None, '2_nD__sn_neighbors': 10, '2_nD__srandom_state': 42}
================================
----------- Scores -----------
2_nD__neighbors_articles_authors : 0.4779
--------- Desc Scores ---------
2_nD__neighbors_articles_authors_det
0 Céline Gicquel : 0.82
1 Loïc Paulevé : 0.73
2 Nikolaus Hansen : 0.69
3 Sébastien Tixeuil : 0.68
4 Philippe Caillou : 0.67
5 Cyril Furtlehner : 0.67
6 Raymond Ros : 0.66
7 Isabelle Guyon : 0.62
8 Dimo Brockhoff : 0.61
9 Paola Tubaro : 0.60
10 Anne Auger : 0.59
11 Yann Ponty : 0.59
12 Sarah Cohen-Boulakia : 0.58
13 Olivier Teytaud : 0.57
14 Nicolas Bredeche : 0.54
15 Marc Baboulin : 0.54
16 Tobias Isenberg : 0.54
17 Albert Cohen : 0.53
18 Guillaume Charpiat : 0.52
19 Chantal Reynaud : 0.51
20 Fatiha Saïs : 0.51
21 Ioana Manolescu : 0.51
22 Sylvie Boldo : 0.49
23 Lonni Besançon : 0.49
24 Nathalie Pernelle : 0.48
25 Franck Cappello : 0.48
26 Petra Isenberg : 0.48
27 Guillaume Melquiond : 0.47
28 Claude Marché : 0.47
29 Marc Schoenauer : 0.44
30 Jean-Daniel Fekete : 0.44
31 François Goasdoué : 0.43
32 Fatiha Zaidi : 0.39
33 Olivier Chapuis : 0.38
34 Michèle Sebag : 0.38
35 Nathann Cohen : 0.38
36 Steven Martin : 0.36
37 Anastasia Bezerianos : 0.32
38 Pierre Dragicevic : 0.32
39 Wendy Mackay : 0.30
40 Michel Beaudouin-Lafon : 0.30
41 Alain Denise : 0.30
42 Balázs Kégl : 0.29
43 Caroline Appert : 0.27
44 Evelyne Lutton : 0.27
45 Johanne Cohen : 0.27
46 Emmanuel Pietriga : 0.26
47 Wendy E. Mackay : 0.20
Name: 0, dtype: object
VALUE : 0.4779
--------- Raw Scores ---------
2_nD__neighbors_articles_authors
Sébastien Tixeuil 0.680851
Michèle Sebag 0.383942
Johanne Cohen 0.272093
Albert Cohen 0.529310
Wendy E. Mackay 0.200000
Philippe Caillou 0.674419
Alain Denise 0.297222
Jean-Daniel Fekete 0.437107
Emmanuel Pietriga 0.260317
Yann Ponty 0.586364
Marc Schoenauer 0.443165
Franck Cappello 0.480488
Caroline Appert 0.273913
Michel Beaudouin-Lafon 0.297531
Wendy Mackay 0.302128
Anne Auger 0.591139
Evelyne Lutton 0.272973
Pierre Dragicevic 0.317284
Ioana Manolescu 0.507317
Nikolaus Hansen 0.686420
Nicolas Bredeche 0.544118
Olivier Teytaud 0.567925
François Goasdoué 0.430189
Nathalie Pernelle 0.482353
Fatiha Saïs 0.507317
Sarah Cohen-Boulakia 0.581818
Claude Marché 0.465957
Chantal Reynaud 0.513333
Olivier Chapuis 0.384615
Steven Martin 0.356410
Fatiha Zaidi 0.390625
Balázs Kégl 0.286842
Paola Tubaro 0.602564
Raymond Ros 0.655882
Cyril Furtlehner 0.666667
Anastasia Bezerianos 0.319403
Sylvie Boldo 0.488571
Guillaume Melquiond 0.472727
Marc Baboulin 0.540000
Dimo Brockhoff 0.610256
Nathann Cohen 0.380488
Petra Isenberg 0.479439
Tobias Isenberg 0.536752
Loïc Paulevé 0.726190
Céline Gicquel 0.818421
Isabelle Guyon 0.623596
Guillaume Charpiat 0.522581
Lonni Besançon 0.487879
Name: 0, dtype: float64
================================
Run params : {}
--------------------------------
Scoring params : {}
================================
----------- Scores -----------
custom_score_lsa : 10.0000
--------- Desc Scores ---------
--------- Raw Scores ---------
custom scores:
None
================================
Run params : {'2_nD__rkey': 'lsa', '2_nD__rnum_dims': 50, '2_nD__rnormalize': True, '2_nD__rextra_param': True}
--------------------------------
Scoring params : {'2_nD__smin_score': 30, '2_nD__srecompute': True, '2_nD__ssample_size': None, '2_nD__sn_neighbors': 10, '2_nD__srandom_state': 42, '2_nD__sfactor': 10}
================================
----------- Scores -----------
2_nD__neighbors_articles_authors : 0.4779
2_nD__custom_score_lsa : 10.0000
--------- Desc Scores ---------
2_nD__neighbors_articles_authors_det
0 Céline Gicquel : 0.82
1 Loïc Paulevé : 0.73
2 Nikolaus Hansen : 0.69
3 Sébastien Tixeuil : 0.68
4 Philippe Caillou : 0.67
5 Cyril Furtlehner : 0.67
6 Raymond Ros : 0.66
7 Isabelle Guyon : 0.62
8 Dimo Brockhoff : 0.61
9 Paola Tubaro : 0.60
10 Anne Auger : 0.59
11 Yann Ponty : 0.59
12 Sarah Cohen-Boulakia : 0.58
13 Olivier Teytaud : 0.57
14 Nicolas Bredeche : 0.54
15 Marc Baboulin : 0.54
16 Tobias Isenberg : 0.54
17 Albert Cohen : 0.53
18 Guillaume Charpiat : 0.52
19 Chantal Reynaud : 0.51
20 Fatiha Saïs : 0.51
21 Ioana Manolescu : 0.51
22 Sylvie Boldo : 0.49
23 Lonni Besançon : 0.49
24 Nathalie Pernelle : 0.48
25 Franck Cappello : 0.48
26 Petra Isenberg : 0.48
27 Guillaume Melquiond : 0.47
28 Claude Marché : 0.47
29 Marc Schoenauer : 0.44
30 Jean-Daniel Fekete : 0.44
31 François Goasdoué : 0.43
32 Fatiha Zaidi : 0.39
33 Olivier Chapuis : 0.38
34 Michèle Sebag : 0.38
35 Nathann Cohen : 0.38
36 Steven Martin : 0.36
37 Anastasia Bezerianos : 0.32
38 Pierre Dragicevic : 0.32
39 Wendy Mackay : 0.30
40 Michel Beaudouin-Lafon : 0.30
41 Alain Denise : 0.30
42 Balázs Kégl : 0.29
43 Caroline Appert : 0.27
44 Evelyne Lutton : 0.27
45 Johanne Cohen : 0.27
46 Emmanuel Pietriga : 0.26
47 Wendy E. Mackay : 0.20
Name: 0, dtype: object
VALUE : 0.4779
--------- Raw Scores ---------
2_nD__neighbors_articles_authors
Sébastien Tixeuil 0.680851
Michèle Sebag 0.383942
Johanne Cohen 0.272093
Albert Cohen 0.529310
Wendy E. Mackay 0.200000
Philippe Caillou 0.674419
Alain Denise 0.297222
Jean-Daniel Fekete 0.437107
Emmanuel Pietriga 0.260317
Yann Ponty 0.586364
Marc Schoenauer 0.443165
Franck Cappello 0.480488
Caroline Appert 0.273913
Michel Beaudouin-Lafon 0.297531
Wendy Mackay 0.302128
Anne Auger 0.591139
Evelyne Lutton 0.272973
Pierre Dragicevic 0.317284
Ioana Manolescu 0.507317
Nikolaus Hansen 0.686420
Nicolas Bredeche 0.544118
Olivier Teytaud 0.567925
François Goasdoué 0.430189
Nathalie Pernelle 0.482353
Fatiha Saïs 0.507317
Sarah Cohen-Boulakia 0.581818
Claude Marché 0.465957
Chantal Reynaud 0.513333
Olivier Chapuis 0.384615
Steven Martin 0.356410
Fatiha Zaidi 0.390625
Balázs Kégl 0.286842
Paola Tubaro 0.602564
Raymond Ros 0.655882
Cyril Furtlehner 0.666667
Anastasia Bezerianos 0.319403
Sylvie Boldo 0.488571
Guillaume Melquiond 0.472727
Marc Baboulin 0.540000
Dimo Brockhoff 0.610256
Nathann Cohen 0.380488
Petra Isenberg 0.479439
Tobias Isenberg 0.536752
Loïc Paulevé 0.726190
Céline Gicquel 0.818421
Isabelle Guyon 0.623596
Guillaume Charpiat 0.522581
Lonni Besançon 0.487879
Name: 0, dtype: float64
2_projection_nD scores:
None
umap matrices generated.
================================
Run params : {'3_2D__rkey': 'umap', '3_2D__rmetric': 'euclidean', '3_2D__rn_neighbors': 20, '3_2D__rmin_dist': 0.25, '3_2D__rinit': 'random', '3_2D__rlearning_rate': 1.0, '3_2D__rn_epochs': None, '3_2D__rrandom_state': None}
--------------------------------
Scoring params : {'3_2D__smin_score': 30, '3_2D__srecompute': True, '3_2D__ssample_size': None, '3_2D__sn_neighbors': 10, '3_2D__srandom_state': 42}
================================
----------- Scores -----------
3_2D__neighbors_articles_authors : 0.4156
3_2D__2_nD_3_2D_neighbors_articles_authors_dim_mean : -0.0623
3_2D__2_nD_3_2D_neighbors_articles_authors_dim_median : -0.0483
--------- Desc Scores ---------
3_2D__neighbors_articles_authors_det
0 Loïc Paulevé : 0.68
1 Céline Gicquel : 0.67
2 Philippe Caillou : 0.60
3 Cyril Furtlehner : 0.60
4 Nikolaus Hansen : 0.59
5 Sébastien Tixeuil : 0.59
6 Yann Ponty : 0.57
7 Tobias Isenberg : 0.52
8 Dimo Brockhoff : 0.52
9 Isabelle Guyon : 0.51
10 Raymond Ros : 0.50
11 Albert Cohen : 0.50
12 Anne Auger : 0.49
13 Olivier Teytaud : 0.49
14 Nathalie Pernelle : 0.49
15 Paola Tubaro : 0.48
16 Guillaume Melquiond : 0.48
17 Franck Cappello : 0.48
18 Nicolas Bredeche : 0.47
19 Marc Baboulin : 0.46
20 Fatiha Saïs : 0.46
21 Sylvie Boldo : 0.45
22 Chantal Reynaud : 0.43
23 Claude Marché : 0.43
24 Ioana Manolescu : 0.42
25 Lonni Besançon : 0.42
26 Sarah Cohen-Boulakia : 0.41
27 Marc Schoenauer : 0.40
28 Jean-Daniel Fekete : 0.39
29 Nathann Cohen : 0.38
30 Steven Martin : 0.37
31 Petra Isenberg : 0.36
32 Michèle Sebag : 0.35
33 Fatiha Zaidi : 0.34
34 Olivier Chapuis : 0.33
35 Guillaume Charpiat : 0.33
36 François Goasdoué : 0.30
37 Anastasia Bezerianos : 0.29
38 Pierre Dragicevic : 0.28
39 Michel Beaudouin-Lafon : 0.28
40 Johanne Cohen : 0.26
41 Alain Denise : 0.25
42 Caroline Appert : 0.24
43 Evelyne Lutton : 0.23
44 Emmanuel Pietriga : 0.22
45 Balázs Kégl : 0.22
46 Wendy Mackay : 0.21
47 Wendy E. Mackay : 0.20
dtype: object
VALUE : 0.4156
3_2D__2_nD_3_2D_neighbors_articles_authors_dim_det
0 Guillaume Charpiat : -0.20
1 Sarah Cohen-Boulakia : -0.17
2 Raymond Ros : -0.15
3 Céline Gicquel : -0.15
4 François Goasdoué : -0.13
5 Paola Tubaro : -0.12
6 Isabelle Guyon : -0.12
7 Petra Isenberg : -0.12
8 Nikolaus Hansen : -0.10
9 Anne Auger : -0.10
10 Dimo Brockhoff : -0.09
11 Sébastien Tixeuil : -0.09
12 Wendy Mackay : -0.09
13 Ioana Manolescu : -0.08
14 Chantal Reynaud : -0.08
15 Olivier Teytaud : -0.08
16 Philippe Caillou : -0.08
17 Marc Baboulin : -0.08
18 Nicolas Bredeche : -0.07
19 Cyril Furtlehner : -0.07
20 Balázs Kégl : -0.07
21 Lonni Besançon : -0.07
22 Olivier Chapuis : -0.05
23 Fatiha Saïs : -0.05
24 Jean-Daniel Fekete : -0.05
25 Fatiha Zaidi : -0.05
26 Marc Schoenauer : -0.05
27 Alain Denise : -0.04
28 Loïc Paulevé : -0.04
29 Sylvie Boldo : -0.04
30 Evelyne Lutton : -0.04
31 Caroline Appert : -0.04
32 Emmanuel Pietriga : -0.04
33 Claude Marché : -0.04
34 Pierre Dragicevic : -0.03
35 Albert Cohen : -0.03
36 Michèle Sebag : -0.03
37 Anastasia Bezerianos : -0.03
38 Tobias Isenberg : -0.02
39 Johanne Cohen : -0.02
40 Michel Beaudouin-Lafon : -0.01
41 Yann Ponty : -0.01
42 Nathann Cohen : -0.00
43 Franck Cappello : -0.00
44 Wendy E. Mackay : 0.00
45 Nathalie Pernelle : 0.00
46 Guillaume Melquiond : 0.01
47 Steven Martin : 0.01
dtype: object
VALUE : -0.0623
--------- Raw Scores ---------
3_2D__neighbors_articles_authors
Sébastien Tixeuil 0.589362
Michèle Sebag 0.352555
Johanne Cohen 0.255814
Albert Cohen 0.496552
Wendy E. Mackay 0.202174
Philippe Caillou 0.597674
Alain Denise 0.252778
Jean-Daniel Fekete 0.389308
Emmanuel Pietriga 0.223810
Yann Ponty 0.574242
Marc Schoenauer 0.397122
Franck Cappello 0.478049
Caroline Appert 0.236957
Michel Beaudouin-Lafon 0.282716
Wendy Mackay 0.212766
Anne Auger 0.494937
Evelyne Lutton 0.232432
Pierre Dragicevic 0.283951
Ioana Manolescu 0.423171
Nikolaus Hansen 0.590123
Nicolas Bredeche 0.473529
Olivier Teytaud 0.488679
François Goasdoué 0.301887
Nathalie Pernelle 0.485294
Fatiha Saïs 0.458537
Sarah Cohen-Boulakia 0.409091
Claude Marché 0.429787
Chantal Reynaud 0.433333
Olivier Chapuis 0.334615
Steven Martin 0.366667
Fatiha Zaidi 0.343750
Balázs Kégl 0.218421
Paola Tubaro 0.482051
Raymond Ros 0.502941
Cyril Furtlehner 0.597436
Anastasia Bezerianos 0.289552
Sylvie Boldo 0.445714
Guillaume Melquiond 0.478788
Marc Baboulin 0.464444
Dimo Brockhoff 0.515385
Nathann Cohen 0.375610
Petra Isenberg 0.363551
Tobias Isenberg 0.517094
Loïc Paulevé 0.683333
Céline Gicquel 0.668421
Isabelle Guyon 0.505618
Guillaume Charpiat 0.325806
Lonni Besançon 0.421212
dtype: float64
3_2D__2_nD_3_2D_neighbors_articles_authors_dim
2_nD__neighbors_articles_authors \
Sébastien Tixeuil 0.680851
Michèle Sebag 0.383942
Johanne Cohen 0.272093
Albert Cohen 0.529310
Wendy E. Mackay 0.200000
Philippe Caillou 0.674419
Alain Denise 0.297222
Jean-Daniel Fekete 0.437107
Emmanuel Pietriga 0.260317
Yann Ponty 0.586364
Marc Schoenauer 0.443165
Franck Cappello 0.480488
Caroline Appert 0.273913
Michel Beaudouin-Lafon 0.297531
Wendy Mackay 0.302128
Anne Auger 0.591139
Evelyne Lutton 0.272973
Pierre Dragicevic 0.317284
Ioana Manolescu 0.507317
Nikolaus Hansen 0.686420
Nicolas Bredeche 0.544118
Olivier Teytaud 0.567925
François Goasdoué 0.430189
Nathalie Pernelle 0.482353
Fatiha Saïs 0.507317
Sarah Cohen-Boulakia 0.581818
Claude Marché 0.465957
Chantal Reynaud 0.513333
Olivier Chapuis 0.384615
Steven Martin 0.356410
Fatiha Zaidi 0.390625
Balázs Kégl 0.286842
Paola Tubaro 0.602564
Raymond Ros 0.655882
Cyril Furtlehner 0.666667
Anastasia Bezerianos 0.319403
Sylvie Boldo 0.488571
Guillaume Melquiond 0.472727
Marc Baboulin 0.540000
Dimo Brockhoff 0.610256
Nathann Cohen 0.380488
Petra Isenberg 0.479439
Tobias Isenberg 0.536752
Loïc Paulevé 0.726190
Céline Gicquel 0.818421
Isabelle Guyon 0.623596
Guillaume Charpiat 0.522581
Lonni Besançon 0.487879
3_2D__neighbors_articles_authors scdec
Sébastien Tixeuil 0.589362 -0.091489
Michèle Sebag 0.352555 -0.031387
Johanne Cohen 0.255814 -0.016279
Albert Cohen 0.496552 -0.032759
Wendy E. Mackay 0.202174 0.002174
Philippe Caillou 0.597674 -0.076744
Alain Denise 0.252778 -0.044444
Jean-Daniel Fekete 0.389308 -0.047799
Emmanuel Pietriga 0.223810 -0.036508
Yann Ponty 0.574242 -0.012121
Marc Schoenauer 0.397122 -0.046043
Franck Cappello 0.478049 -0.002439
Caroline Appert 0.236957 -0.036957
Michel Beaudouin-Lafon 0.282716 -0.014815
Wendy Mackay 0.212766 -0.089362
Anne Auger 0.494937 -0.096203
Evelyne Lutton 0.232432 -0.040541
Pierre Dragicevic 0.283951 -0.033333
Ioana Manolescu 0.423171 -0.084146
Nikolaus Hansen 0.590123 -0.096296
Nicolas Bredeche 0.473529 -0.070588
Olivier Teytaud 0.488679 -0.079245
François Goasdoué 0.301887 -0.128302
Nathalie Pernelle 0.485294 0.002941
Fatiha Saïs 0.458537 -0.048780
Sarah Cohen-Boulakia 0.409091 -0.172727
Claude Marché 0.429787 -0.036170
Chantal Reynaud 0.433333 -0.080000
Olivier Chapuis 0.334615 -0.050000
Steven Martin 0.366667 0.010256
Fatiha Zaidi 0.343750 -0.046875
Balázs Kégl 0.218421 -0.068421
Paola Tubaro 0.482051 -0.120513
Raymond Ros 0.502941 -0.152941
Cyril Furtlehner 0.597436 -0.069231
Anastasia Bezerianos 0.289552 -0.029851
Sylvie Boldo 0.445714 -0.042857
Guillaume Melquiond 0.478788 0.006061
Marc Baboulin 0.464444 -0.075556
Dimo Brockhoff 0.515385 -0.094872
Nathann Cohen 0.375610 -0.004878
Petra Isenberg 0.363551 -0.115888
Tobias Isenberg 0.517094 -0.019658
Loïc Paulevé 0.683333 -0.042857
Céline Gicquel 0.668421 -0.150000
Isabelle Guyon 0.505618 -0.117978
Guillaume Charpiat 0.325806 -0.196774
Lonni Besançon 0.421212 -0.066667
3_projection_2D scores:
None
Nothing in cache, initial Fitting with min_cluster_size=15 Found 68 clusters in 0.28237541199996485s
Max Fitting with min_cluster_size=30 Found 47 clusters in 0.10517458799949964s
Max Fitting with min_cluster_size=60 Found 23 clusters in 0.10227592600131175s
Max Fitting with min_cluster_size=120 Found 8 clusters in 0.09809423800106742s
No need Re-Fitting with min_cluster_size=120
Clusters cached: [8, 23, 47, 68]
Nothing in cache, initial Fitting with min_cluster_size=15 Found 68 clusters in 0.09952020899800118s
Max Fitting with min_cluster_size=30 Found 47 clusters in 0.10086908200173639s
Max Fitting with min_cluster_size=60 Found 23 clusters in 0.10139435800010688s
Midpoint Fitting with min_cluster_size=45 Found 29 clusters in 0.10205801800111658s
Midpoint Fitting with min_cluster_size=52 Found 26 clusters in 0.09973800700026914s
No need Re-Fitting with min_cluster_size=52
Clusters cached: [23, 26, 29, 47, 68]
================================
Run params : {'4_clus__rkey': 'hdbscan', '4_clus__rbase_factor': 3}
--------------------------------
Scoring params : {'4_clus__siter_stab': 2, '4_clus__sremove_stab': [0, 0.01, 0.03, 0.1, 0.25], '4_clus__smetric': 'euclidean', '4_clus__srandom_state': None}
================================
----------- Scores -----------
4_clus__nb_clust_0 : 8.0000
4_clus__silhouette_0 : 0.0580
4_clus__avg_word_couv_0 : 0.4744
4_clus__med_word_couv_0 : 0.5177
4_clus__avg_word_couv_minus_0 : 0.4455
4_clus__big_small_ratio_0 : 8.8770
4_clus__stab_clus_0 : 0.1875
4_clus__nb_clust_1 : 24.0000
4_clus__silhouette_1 : 0.1464
4_clus__avg_word_couv_1 : 0.6585
4_clus__med_word_couv_1 : 0.6773
4_clus__avg_word_couv_minus_1 : 0.6338
4_clus__big_small_ratio_1 : 19.2000
4_clus__stab_clus_1 : 0.1269
4_clus__avg_stab_avg : 0.1572
4_clus__avg_couv_avg : 0.5664
4_clus__clu_score : 0.3618
--------- Desc Scores ---------
4_clus__clus_eval_pos_0_det
0 networking internet architecture, neural netwo...
1 visualizations, interfaces : s 720 stb 0.80 + ...
2 query, visual analytics : s 554 stb 0.00 + 0.3...
3 quantitative methods, bioinformatics : s 283 s...
4 logic science, semantics : s 222 stb 0.10 + 0....
5 architectures, programming languages : s 188 s...
6 services, cluster computing : s 125 stb 0.00 +...
7 fluid, numerical simulations : s 122 stb 0.00 ...
dtype: object
VALUE : 0.4744
4_clus__clus_eval_pos_1_det
0 interaction techniques, virtual : s 720 stb 0....
1 ontology, documents : s 185 stb 0.70 + 0.57 - ...
2 image, clustering : s 128 stb 0.00 + 0.52 - 0.04
3 internet architecture, energy : s 128 stb 0.00...
4 vertices, minimum : s 126 stb 0.00 + 0.67 - 0.03
5 service, fault : s 125 stb 0.00 + 0.68 - 0.04
6 secondary structure, protein : s 123 stb 0.00 ...
7 dynamics, mechanics : s 122 stb 0.00 + 0.62 - ...
8 multi objective, belief propagation : s 122 st...
9 compiler, optimizations : s 120 stb 0.00 + 0.5...
10 humanities social sciences, social networks : ...
11 biology, metabolic : s 114 stb 0.10 + 0.54 - 0.02
12 adaptive, multi armed : s 108 stb 0.20 + 0.63 ...
13 covariance matrix adaptation, benchmarking : s...
14 queries : s 99 stb 0.00 + 0.76 - 0.02
15 verification : s 85 stb 0.00 + 0.85 - 0.03
16 floating point, solver : s 79 stb 0.30 + 0.81 ...
17 combinatorics, trees : s 78 stb 0.00 + 0.67 - ...
18 monte carlo, games : s 74 stb 0.20 + 0.97 - 0.02
19 calculus, object oriented : s 71 stb 0.00 + 0....
20 automata, boolean networks : s 71 stb 0.10 + 0...
21 challenge, reinforcement learning : s 65 stb 0...
22 analytics : s 63 stb 0.00 + 0.95 - 0.01
23 stabilizing, population protocols : s 62 stb 0...
24 linear systems, matrices : s 61 stb 0.00 + 0.7...
25 optimization control, bounds : s 55 stb 0.10 +...
dtype: object
VALUE : 0.6585
--------- Raw Scores ---------
4_clustering scores:
None
['2_nD__neighbors_articles_authors', '3_2D__neighbors_articles_authors', '4_clus__clu_score']
================================
Run params : {}
--------------------------------
Scoring params : {'6_pst__sname_list': ['2_nD__neighbors_articles_authors', '3_2D__neighbors_articles_authors', '4_clus__clu_score']}
================================
----------- Scores -----------
6_pst__final_score : 0.4184
--------- Desc Scores ---------
6_pst__final_score_det
0 2_nD__neighbors_articles_authors : 0.48
1 3_2D__neighbors_articles_authors : 0.42
2 4_clus__clu_score : 0.36
dtype: object
VALUE : 0.4184
--------- Raw Scores ---------
6_post scores:
None
Finished running step!
Running experiment for parameters:
{'id': '014dbb63a341d2622f25', 'robustseed': 0, 'authors__filter_min_score': 4, 'filter_min_score': 6, 'projection_nD': {'key': 'lsa', 'num_dims': 50, 'extra_param': True}, 'projection_2D': {'key': 'umap', 'n_neighbors': 20, 'min_dist': 0.5, 'metric': 'euclidean'}}
Matrices of type mat for articles, authors, teams, labs, words already exist. Skipping creation.
Matrices of type lsa for articles, authors, teams, labs, words already exist. Skipping creation.
================================
Run params : {'2_nD__rkey': 'lsa', '2_nD__rnum_dims': 50, '2_nD__rnormalize': True, '2_nD__rextra_param': True}
--------------------------------
Scoring params : {'2_nD__smin_score': 30, '2_nD__srecompute': True, '2_nD__ssample_size': None, '2_nD__sn_neighbors': 10, '2_nD__srandom_state': 42}
================================
----------- Scores -----------
2_nD__neighbors_articles_authors : 0.4779
--------- Desc Scores ---------
2_nD__neighbors_articles_authors_det
0 Céline Gicquel : 0.82
1 Loïc Paulevé : 0.73
2 Nikolaus Hansen : 0.69
3 Sébastien Tixeuil : 0.68
4 Philippe Caillou : 0.67
5 Cyril Furtlehner : 0.67
6 Raymond Ros : 0.66
7 Isabelle Guyon : 0.62
8 Dimo Brockhoff : 0.61
9 Paola Tubaro : 0.60
10 Anne Auger : 0.59
11 Yann Ponty : 0.59
12 Sarah Cohen-Boulakia : 0.58
13 Olivier Teytaud : 0.57
14 Nicolas Bredeche : 0.54
15 Marc Baboulin : 0.54
16 Tobias Isenberg : 0.54
17 Albert Cohen : 0.53
18 Guillaume Charpiat : 0.52
19 Chantal Reynaud : 0.51
20 Fatiha Saïs : 0.51
21 Ioana Manolescu : 0.51
22 Sylvie Boldo : 0.49
23 Lonni Besançon : 0.49
24 Nathalie Pernelle : 0.48
25 Franck Cappello : 0.48
26 Petra Isenberg : 0.48
27 Guillaume Melquiond : 0.47
28 Claude Marché : 0.47
29 Marc Schoenauer : 0.44
30 Jean-Daniel Fekete : 0.44
31 François Goasdoué : 0.43
32 Fatiha Zaidi : 0.39
33 Olivier Chapuis : 0.38
34 Michèle Sebag : 0.38
35 Nathann Cohen : 0.38
36 Steven Martin : 0.36
37 Anastasia Bezerianos : 0.32
38 Pierre Dragicevic : 0.32
39 Wendy Mackay : 0.30
40 Michel Beaudouin-Lafon : 0.30
41 Alain Denise : 0.30
42 Balázs Kégl : 0.29
43 Caroline Appert : 0.27
44 Evelyne Lutton : 0.27
45 Johanne Cohen : 0.27
46 Emmanuel Pietriga : 0.26
47 Wendy E. Mackay : 0.20
Name: 0, dtype: object
VALUE : 0.4779
--------- Raw Scores ---------
2_nD__neighbors_articles_authors
Sébastien Tixeuil 0.680851
Michèle Sebag 0.383942
Johanne Cohen 0.272093
Albert Cohen 0.529310
Wendy E. Mackay 0.200000
Philippe Caillou 0.674419
Alain Denise 0.297222
Jean-Daniel Fekete 0.437107
Emmanuel Pietriga 0.260317
Yann Ponty 0.586364
Marc Schoenauer 0.443165
Franck Cappello 0.480488
Caroline Appert 0.273913
Michel Beaudouin-Lafon 0.297531
Wendy Mackay 0.302128
Anne Auger 0.591139
Evelyne Lutton 0.272973
Pierre Dragicevic 0.317284
Ioana Manolescu 0.507317
Nikolaus Hansen 0.686420
Nicolas Bredeche 0.544118
Olivier Teytaud 0.567925
François Goasdoué 0.430189
Nathalie Pernelle 0.482353
Fatiha Saïs 0.507317
Sarah Cohen-Boulakia 0.581818
Claude Marché 0.465957
Chantal Reynaud 0.513333
Olivier Chapuis 0.384615
Steven Martin 0.356410
Fatiha Zaidi 0.390625
Balázs Kégl 0.286842
Paola Tubaro 0.602564
Raymond Ros 0.655882
Cyril Furtlehner 0.666667
Anastasia Bezerianos 0.319403
Sylvie Boldo 0.488571
Guillaume Melquiond 0.472727
Marc Baboulin 0.540000
Dimo Brockhoff 0.610256
Nathann Cohen 0.380488
Petra Isenberg 0.479439
Tobias Isenberg 0.536752
Loïc Paulevé 0.726190
Céline Gicquel 0.818421
Isabelle Guyon 0.623596
Guillaume Charpiat 0.522581
Lonni Besançon 0.487879
Name: 0, dtype: float64
================================
Run params : {}
--------------------------------
Scoring params : {}
================================
----------- Scores -----------
custom_score_lsa : 10.0000
--------- Desc Scores ---------
--------- Raw Scores ---------
custom scores:
None
================================
Run params : {'2_nD__rkey': 'lsa', '2_nD__rnum_dims': 50, '2_nD__rnormalize': True, '2_nD__rextra_param': True}
--------------------------------
Scoring params : {'2_nD__smin_score': 30, '2_nD__srecompute': True, '2_nD__ssample_size': None, '2_nD__sn_neighbors': 10, '2_nD__srandom_state': 42, '2_nD__sfactor': 10}
================================
----------- Scores -----------
2_nD__neighbors_articles_authors : 0.4779
2_nD__custom_score_lsa : 10.0000
--------- Desc Scores ---------
2_nD__neighbors_articles_authors_det
0 Céline Gicquel : 0.82
1 Loïc Paulevé : 0.73
2 Nikolaus Hansen : 0.69
3 Sébastien Tixeuil : 0.68
4 Philippe Caillou : 0.67
5 Cyril Furtlehner : 0.67
6 Raymond Ros : 0.66
7 Isabelle Guyon : 0.62
8 Dimo Brockhoff : 0.61
9 Paola Tubaro : 0.60
10 Anne Auger : 0.59
11 Yann Ponty : 0.59
12 Sarah Cohen-Boulakia : 0.58
13 Olivier Teytaud : 0.57
14 Nicolas Bredeche : 0.54
15 Marc Baboulin : 0.54
16 Tobias Isenberg : 0.54
17 Albert Cohen : 0.53
18 Guillaume Charpiat : 0.52
19 Chantal Reynaud : 0.51
20 Fatiha Saïs : 0.51
21 Ioana Manolescu : 0.51
22 Sylvie Boldo : 0.49
23 Lonni Besançon : 0.49
24 Nathalie Pernelle : 0.48
25 Franck Cappello : 0.48
26 Petra Isenberg : 0.48
27 Guillaume Melquiond : 0.47
28 Claude Marché : 0.47
29 Marc Schoenauer : 0.44
30 Jean-Daniel Fekete : 0.44
31 François Goasdoué : 0.43
32 Fatiha Zaidi : 0.39
33 Olivier Chapuis : 0.38
34 Michèle Sebag : 0.38
35 Nathann Cohen : 0.38
36 Steven Martin : 0.36
37 Anastasia Bezerianos : 0.32
38 Pierre Dragicevic : 0.32
39 Wendy Mackay : 0.30
40 Michel Beaudouin-Lafon : 0.30
41 Alain Denise : 0.30
42 Balázs Kégl : 0.29
43 Caroline Appert : 0.27
44 Evelyne Lutton : 0.27
45 Johanne Cohen : 0.27
46 Emmanuel Pietriga : 0.26
47 Wendy E. Mackay : 0.20
Name: 0, dtype: object
VALUE : 0.4779
--------- Raw Scores ---------
2_nD__neighbors_articles_authors
Sébastien Tixeuil 0.680851
Michèle Sebag 0.383942
Johanne Cohen 0.272093
Albert Cohen 0.529310
Wendy E. Mackay 0.200000
Philippe Caillou 0.674419
Alain Denise 0.297222
Jean-Daniel Fekete 0.437107
Emmanuel Pietriga 0.260317
Yann Ponty 0.586364
Marc Schoenauer 0.443165
Franck Cappello 0.480488
Caroline Appert 0.273913
Michel Beaudouin-Lafon 0.297531
Wendy Mackay 0.302128
Anne Auger 0.591139
Evelyne Lutton 0.272973
Pierre Dragicevic 0.317284
Ioana Manolescu 0.507317
Nikolaus Hansen 0.686420
Nicolas Bredeche 0.544118
Olivier Teytaud 0.567925
François Goasdoué 0.430189
Nathalie Pernelle 0.482353
Fatiha Saïs 0.507317
Sarah Cohen-Boulakia 0.581818
Claude Marché 0.465957
Chantal Reynaud 0.513333
Olivier Chapuis 0.384615
Steven Martin 0.356410
Fatiha Zaidi 0.390625
Balázs Kégl 0.286842
Paola Tubaro 0.602564
Raymond Ros 0.655882
Cyril Furtlehner 0.666667
Anastasia Bezerianos 0.319403
Sylvie Boldo 0.488571
Guillaume Melquiond 0.472727
Marc Baboulin 0.540000
Dimo Brockhoff 0.610256
Nathann Cohen 0.380488
Petra Isenberg 0.479439
Tobias Isenberg 0.536752
Loïc Paulevé 0.726190
Céline Gicquel 0.818421
Isabelle Guyon 0.623596
Guillaume Charpiat 0.522581
Lonni Besançon 0.487879
Name: 0, dtype: float64
2_projection_nD scores:
None
umap matrices generated.
================================
Run params : {'3_2D__rkey': 'umap', '3_2D__rmetric': 'euclidean', '3_2D__rn_neighbors': 20, '3_2D__rmin_dist': 0.5, '3_2D__rinit': 'random', '3_2D__rlearning_rate': 1.0, '3_2D__rn_epochs': None, '3_2D__rrandom_state': None}
--------------------------------
Scoring params : {'3_2D__smin_score': 30, '3_2D__srecompute': True, '3_2D__ssample_size': None, '3_2D__sn_neighbors': 10, '3_2D__srandom_state': 42}
================================
----------- Scores -----------
3_2D__neighbors_articles_authors : 0.3938
3_2D__2_nD_3_2D_neighbors_articles_authors_dim_mean : -0.0840
3_2D__2_nD_3_2D_neighbors_articles_authors_dim_median : -0.0802
--------- Desc Scores ---------
3_2D__neighbors_articles_authors_det
0 Céline Gicquel : 0.66
1 Loïc Paulevé : 0.64
2 Sébastien Tixeuil : 0.59
3 Raymond Ros : 0.56
4 Nikolaus Hansen : 0.55
5 Philippe Caillou : 0.54
6 Olivier Teytaud : 0.52
7 Cyril Furtlehner : 0.51
8 Guillaume Melquiond : 0.51
9 Yann Ponty : 0.51
10 Nicolas Bredeche : 0.50
11 Dimo Brockhoff : 0.50
12 Albert Cohen : 0.48
13 Anne Auger : 0.47
14 Sarah Cohen-Boulakia : 0.45
15 Isabelle Guyon : 0.44
16 Ioana Manolescu : 0.43
17 Marc Baboulin : 0.42
18 Tobias Isenberg : 0.42
19 Fatiha Saïs : 0.42
20 Franck Cappello : 0.41
21 Jean-Daniel Fekete : 0.41
22 Chantal Reynaud : 0.41
23 Sylvie Boldo : 0.41
24 Claude Marché : 0.39
25 Paola Tubaro : 0.38
26 Marc Schoenauer : 0.38
27 Lonni Besançon : 0.38
28 Petra Isenberg : 0.36
29 Steven Martin : 0.35
30 Nathann Cohen : 0.35
31 Guillaume Charpiat : 0.35
32 François Goasdoué : 0.35
33 Nathalie Pernelle : 0.34
34 Michèle Sebag : 0.32
35 Olivier Chapuis : 0.32
36 Fatiha Zaidi : 0.30
37 Alain Denise : 0.27
38 Pierre Dragicevic : 0.27
39 Michel Beaudouin-Lafon : 0.26
40 Anastasia Bezerianos : 0.26
41 Caroline Appert : 0.25
42 Johanne Cohen : 0.23
43 Balázs Kégl : 0.22
44 Wendy Mackay : 0.22
45 Emmanuel Pietriga : 0.21
46 Evelyne Lutton : 0.18
47 Wendy E. Mackay : 0.18
dtype: object
VALUE : 0.3938
3_2D__2_nD_3_2D_neighbors_articles_authors_dim_det
0 Paola Tubaro : -0.22
1 Isabelle Guyon : -0.18
2 Guillaume Charpiat : -0.17
3 Céline Gicquel : -0.16
4 Cyril Furtlehner : -0.15
5 Nathalie Pernelle : -0.14
6 Philippe Caillou : -0.14
7 Sarah Cohen-Boulakia : -0.14
8 Nikolaus Hansen : -0.13
9 Petra Isenberg : -0.12
10 Anne Auger : -0.12
11 Tobias Isenberg : -0.12
12 Marc Baboulin : -0.12
13 Lonni Besançon : -0.11
14 Dimo Brockhoff : -0.11
15 Chantal Reynaud : -0.10
16 Raymond Ros : -0.10
17 Fatiha Zaidi : -0.09
18 Fatiha Saïs : -0.09
19 Sébastien Tixeuil : -0.09
20 Evelyne Lutton : -0.09
21 Loïc Paulevé : -0.09
22 François Goasdoué : -0.08
23 Ioana Manolescu : -0.08
24 Sylvie Boldo : -0.08
25 Wendy Mackay : -0.08
26 Yann Ponty : -0.08
27 Claude Marché : -0.07
28 Franck Cappello : -0.07
29 Olivier Chapuis : -0.06
30 Marc Schoenauer : -0.06
31 Balázs Kégl : -0.06
32 Michèle Sebag : -0.06
33 Anastasia Bezerianos : -0.06
34 Emmanuel Pietriga : -0.05
35 Albert Cohen : -0.05
36 Pierre Dragicevic : -0.05
37 Olivier Teytaud : -0.05
38 Nicolas Bredeche : -0.04
39 Johanne Cohen : -0.04
40 Michel Beaudouin-Lafon : -0.04
41 Nathann Cohen : -0.03
42 Jean-Daniel Fekete : -0.03
43 Alain Denise : -0.03
44 Wendy E. Mackay : -0.02
45 Caroline Appert : -0.02
46 Steven Martin : -0.00
47 Guillaume Melquiond : 0.04
dtype: object
VALUE : -0.0840
--------- Raw Scores ---------
3_2D__neighbors_articles_authors
Sébastien Tixeuil 0.591489
Michèle Sebag 0.323358
Johanne Cohen 0.234884
Albert Cohen 0.477586
Wendy E. Mackay 0.178261
Philippe Caillou 0.537209
Alain Denise 0.272222
Jean-Daniel Fekete 0.411321
Emmanuel Pietriga 0.206349
Yann Ponty 0.509091
Marc Schoenauer 0.379856
Franck Cappello 0.412195
Caroline Appert 0.254348
Michel Beaudouin-Lafon 0.260494
Wendy Mackay 0.223404
Anne Auger 0.470886
Evelyne Lutton 0.183784
Pierre Dragicevic 0.266667
Ioana Manolescu 0.426829
Nikolaus Hansen 0.551852
Nicolas Bredeche 0.502941
Olivier Teytaud 0.520755
François Goasdoué 0.345283
Nathalie Pernelle 0.344118
Fatiha Saïs 0.417073
Sarah Cohen-Boulakia 0.445455
Claude Marché 0.393617
Chantal Reynaud 0.410000
Olivier Chapuis 0.321154
Steven Martin 0.353846
Fatiha Zaidi 0.300000
Balázs Kégl 0.223684
Paola Tubaro 0.382051
Raymond Ros 0.558824
Cyril Furtlehner 0.512821
Anastasia Bezerianos 0.259701
Sylvie Boldo 0.408571
Guillaume Melquiond 0.509091
Marc Baboulin 0.422222
Dimo Brockhoff 0.502564
Nathann Cohen 0.348780
Petra Isenberg 0.357009
Tobias Isenberg 0.418803
Loïc Paulevé 0.640476
Céline Gicquel 0.663158
Isabelle Guyon 0.444944
Guillaume Charpiat 0.348387
Lonni Besançon 0.375758
dtype: float64
3_2D__2_nD_3_2D_neighbors_articles_authors_dim
2_nD__neighbors_articles_authors \
Sébastien Tixeuil 0.680851
Michèle Sebag 0.383942
Johanne Cohen 0.272093
Albert Cohen 0.529310
Wendy E. Mackay 0.200000
Philippe Caillou 0.674419
Alain Denise 0.297222
Jean-Daniel Fekete 0.437107
Emmanuel Pietriga 0.260317
Yann Ponty 0.586364
Marc Schoenauer 0.443165
Franck Cappello 0.480488
Caroline Appert 0.273913
Michel Beaudouin-Lafon 0.297531
Wendy Mackay 0.302128
Anne Auger 0.591139
Evelyne Lutton 0.272973
Pierre Dragicevic 0.317284
Ioana Manolescu 0.507317
Nikolaus Hansen 0.686420
Nicolas Bredeche 0.544118
Olivier Teytaud 0.567925
François Goasdoué 0.430189
Nathalie Pernelle 0.482353
Fatiha Saïs 0.507317
Sarah Cohen-Boulakia 0.581818
Claude Marché 0.465957
Chantal Reynaud 0.513333
Olivier Chapuis 0.384615
Steven Martin 0.356410
Fatiha Zaidi 0.390625
Balázs Kégl 0.286842
Paola Tubaro 0.602564
Raymond Ros 0.655882
Cyril Furtlehner 0.666667
Anastasia Bezerianos 0.319403
Sylvie Boldo 0.488571
Guillaume Melquiond 0.472727
Marc Baboulin 0.540000
Dimo Brockhoff 0.610256
Nathann Cohen 0.380488
Petra Isenberg 0.479439
Tobias Isenberg 0.536752
Loïc Paulevé 0.726190
Céline Gicquel 0.818421
Isabelle Guyon 0.623596
Guillaume Charpiat 0.522581
Lonni Besançon 0.487879
3_2D__neighbors_articles_authors scdec
Sébastien Tixeuil 0.591489 -0.089362
Michèle Sebag 0.323358 -0.060584
Johanne Cohen 0.234884 -0.037209
Albert Cohen 0.477586 -0.051724
Wendy E. Mackay 0.178261 -0.021739
Philippe Caillou 0.537209 -0.137209
Alain Denise 0.272222 -0.025000
Jean-Daniel Fekete 0.411321 -0.025786
Emmanuel Pietriga 0.206349 -0.053968
Yann Ponty 0.509091 -0.077273
Marc Schoenauer 0.379856 -0.063309
Franck Cappello 0.412195 -0.068293
Caroline Appert 0.254348 -0.019565
Michel Beaudouin-Lafon 0.260494 -0.037037
Wendy Mackay 0.223404 -0.078723
Anne Auger 0.470886 -0.120253
Evelyne Lutton 0.183784 -0.089189
Pierre Dragicevic 0.266667 -0.050617
Ioana Manolescu 0.426829 -0.080488
Nikolaus Hansen 0.551852 -0.134568
Nicolas Bredeche 0.502941 -0.041176
Olivier Teytaud 0.520755 -0.047170
François Goasdoué 0.345283 -0.084906
Nathalie Pernelle 0.344118 -0.138235
Fatiha Saïs 0.417073 -0.090244
Sarah Cohen-Boulakia 0.445455 -0.136364
Claude Marché 0.393617 -0.072340
Chantal Reynaud 0.410000 -0.103333
Olivier Chapuis 0.321154 -0.063462
Steven Martin 0.353846 -0.002564
Fatiha Zaidi 0.300000 -0.090625
Balázs Kégl 0.223684 -0.063158
Paola Tubaro 0.382051 -0.220513
Raymond Ros 0.558824 -0.097059
Cyril Furtlehner 0.512821 -0.153846
Anastasia Bezerianos 0.259701 -0.059701
Sylvie Boldo 0.408571 -0.080000
Guillaume Melquiond 0.509091 0.036364
Marc Baboulin 0.422222 -0.117778
Dimo Brockhoff 0.502564 -0.107692
Nathann Cohen 0.348780 -0.031707
Petra Isenberg 0.357009 -0.122430
Tobias Isenberg 0.418803 -0.117949
Loïc Paulevé 0.640476 -0.085714
Céline Gicquel 0.663158 -0.155263
Isabelle Guyon 0.444944 -0.178652
Guillaume Charpiat 0.348387 -0.174194
Lonni Besançon 0.375758 -0.112121
3_projection_2D scores:
None
Nothing in cache, initial Fitting with min_cluster_size=15 Found 56 clusters in 0.286107854997681s
Max Fitting with min_cluster_size=30 Found 39 clusters in 0.10574022000218974s
Max Fitting with min_cluster_size=60 Found 4 clusters in 0.10681544199906057s
Midpoint Fitting with min_cluster_size=45 Found 2 clusters in 0.10889044299983652s
Midpoint Fitting with min_cluster_size=37 Found 2 clusters in 0.11201030500160414s
Re-Fitting with min_cluster_size=30 Found 39 clusters in 0.10461806199964485s
Clusters cached: [2, 2, 4, 39, 56]
Nothing in cache, initial Fitting with min_cluster_size=15 Found 56 clusters in 0.10496335600328166s
Max Fitting with min_cluster_size=30 Found 39 clusters in 0.10403247299836949s
Max Fitting with min_cluster_size=60 Found 4 clusters in 0.10714149600244127s
Midpoint Fitting with min_cluster_size=45 Found 2 clusters in 0.11016929000106757s
Midpoint Fitting with min_cluster_size=37 Found 2 clusters in 0.11238704099741881s
Re-Fitting with min_cluster_size=30 Found 39 clusters in 0.10525423400031286s
Clusters cached: [2, 2, 4, 39, 56]
================================
Run params : {'4_clus__rkey': 'hdbscan', '4_clus__rbase_factor': 3}
--------------------------------
Scoring params : {'4_clus__siter_stab': 2, '4_clus__sremove_stab': [0, 0.01, 0.03, 0.1, 0.25], '4_clus__smetric': 'euclidean', '4_clus__srandom_state': None}
================================
----------- Scores -----------
4_clus__nb_clust_0 : 8.0000
4_clus__silhouette_0 : 0.0406
4_clus__avg_word_couv_0 : 0.7675
4_clus__med_word_couv_0 : 0.7551
4_clus__avg_word_couv_minus_0 : 0.7440
4_clus__big_small_ratio_0 : 49.7000
4_clus__stab_clus_0 : 0.0000
4_clus__nb_clust_1 : 24.0000
4_clus__silhouette_1 : 0.0406
4_clus__avg_word_couv_1 : 0.6261
4_clus__med_word_couv_1 : 0.6250
4_clus__avg_word_couv_minus_1 : 0.6076
4_clus__big_small_ratio_1 : 49.7000
4_clus__stab_clus_1 : 0.0000
4_clus__avg_stab_avg : 0.0000
4_clus__avg_couv_avg : 0.6968
4_clus__clu_score : 0.3484
--------- Desc Scores ---------
4_clus__clus_eval_pos_0_det
0 ontology, humanities social sciences : s 311 s...
1 semantics, architectures : s 164 stb 0.00 + 0....
2 vertices, minimum : s 136 stb 0.00 + 0.66 - 0.02
3 biology, biological : s 130 stb 0.00 + 0.58 - ...
4 image, natural language : s 130 stb 0.00 + 0.5...
5 inference, gradient : s 125 stb 0.00 + 0.44 - ...
6 networking internet architecture, packet : s 1...
7 fluid, numerical simulations : s 95 stb 0.00 +...
8 logic science, verification : s 80 stb 0.00 + ...
9 visualizations : s 79 stb 0.00 + 0.85 - 0.02
10 secondary structure, sequence : s 77 stb 0.00 ...
11 query : s 73 stb 0.00 + 0.99 - 0.03
12 monte carlo, games : s 72 stb 0.00 + 0.97 - 0.02
13 combinatorics, trees : s 67 stb 0.00 + 0.72 - ...
14 robots, robotics : s 65 stb 0.00 + 0.92 - 0.01
15 covariance matrix adaptation evolution, matric...
16 fault, stabilizing algorithm : s 64 stb 0.00 +...
17 adaptive, bitcoin : s 62 stb 0.00 + 0.63 - 0.03
18 services, cloud : s 57 stb 0.00 + 0.86 - 0.04
19 creative, music : s 54 stb 0.00 + 0.63 - 0.01
20 visual analytics : s 54 stb 0.00 + 0.85 - 0.01
21 noisy optimization, bounds : s 49 stb 0.00 + 0...
22 floating point : s 49 stb 0.00 + 0.86 - 0.00
23 agent : s 48 stb 0.00 + 0.96 - 0.01
24 testing : s 43 stb 0.00 + 0.93 - 0.02
25 multi objective, planning : s 42 stb 0.00 + 0....
26 scientific, provenance : s 42 stb 0.00 + 0.90 ...
27 movement, navigation techniques : s 41 stb 0.0...
28 challenge, alignment : s 40 stb 0.00 + 0.78 - ...
29 benchmarking : s 40 stb 0.00 + 0.88 - 0.01
30 tangible, spatial : s 40 stb 0.00 + 0.75 - 0.04
31 protein : s 39 stb 0.00 + 0.97 - 0.01
32 automata : s 38 stb 0.00 + 0.76 - 0.01
33 multi armed : s 37 stb 0.00 + 0.78 - 0.00
34 quantum : s 35 stb 0.00 + 0.89 - 0.00
35 mixed integer linear, operations : s 34 stb 0....
36 discrete event systems, diagnosis : s 31 stb 0...
37 linear systems, factorization : s 31 stb 0.00 ...
38 materialized views, xquery : s 30 stb 0.00 + 0...
dtype: object
VALUE : 0.7675
4_clus__clus_eval_pos_1_det
0 ontologies, documents : s 311 stb 0.00 + 0.33 ...
1 compiler, optimizations : s 164 stb 0.00 + 0.4...
2 discrete mathematics, vertex : s 136 stb 0.00 ...
3 image processing, natural language processing ...
4 molecular, metabolic : s 130 stb 0.00 + 0.45 -...
5 statistical, bayesian : s 125 stb 0.00 + 0.43 ...
6 internet architecture, wireless : s 102 stb 0....
7 mechanics, nonlinear : s 95 stb 0.00 + 0.59 - ...
8 program verification, programs : s 80 stb 0.00...
9 information visualization, perception : s 79 s...
10 secondary, sampling : s 77 stb 0.00 + 0.78 - 0.03
11 queries, query answers : s 73 stb 0.00 + 0.74 ...
12 monte carlo search, carlo : s 72 stb 0.00 + 0....
13 permutations, lattice : s 67 stb 0.00 + 0.52 -...
14 covariance matrix adaptation, dimension search...
15 robot, mobile robots : s 65 stb 0.00 + 0.83 - ...
16 message, stabilization : s 64 stb 0.00 + 0.67 ...
17 adaptation, mining : s 62 stb 0.00 + 0.45 - 0.05
18 service, cloud computing : s 57 stb 0.00 + 0.7...
19 analytics : s 54 stb 0.00 + 0.96 - 0.01
20 creativity, musical : s 54 stb 0.00 + 0.57 - 0.01
21 floating point arithmetic, arithmetic : s 49 s...
22 noisy, lower bounds : s 49 stb 0.00 + 0.71 - 0.02
23 multi agent, modeling simulation : s 48 stb 0....
24 conformance, symbolic : s 43 stb 0.00 + 0.58 -...
25 scientific workflows, workflow : s 42 stb 0.00...
26 pareto, divide evolve : s 42 stb 0.00 + 0.64 -...
27 dance, perspective : s 41 stb 0.00 + 0.39 - 0.02
28 automl, competition : s 40 stb 0.00 + 0.62 - 0.01
29 black optimization, testbed : s 40 stb 0.00 + ...
30 interface, scientific visualization : s 40 stb...
31 protein protein, structural biology : s 39 stb...
32 cellular automata, words : s 38 stb 0.00 + 0.5...
33 multi armed bandit, multi armed bandits : s 37...
34 quantum physics, circuits : s 35 stb 0.00 + 0....
35 mixed integer, chance : s 34 stb 0.00 + 0.68 -...
36 dense linear, pivoting : s 31 stb 0.00 + 0.71 ...
37 discrete event, faults : s 31 stb 0.00 + 0.84 ...
38 views, efficiency scalability : s 30 stb 0.00 ...
dtype: object
VALUE : 0.6261
--------- Raw Scores ---------
4_clustering scores:
None
['2_nD__neighbors_articles_authors', '3_2D__neighbors_articles_authors', '4_clus__clu_score']
================================
Run params : {}
--------------------------------
Scoring params : {'6_pst__sname_list': ['2_nD__neighbors_articles_authors', '3_2D__neighbors_articles_authors', '4_clus__clu_score']}
================================
----------- Scores -----------
6_pst__final_score : 0.4067
--------- Desc Scores ---------
6_pst__final_score_det
0 2_nD__neighbors_articles_authors : 0.48
1 3_2D__neighbors_articles_authors : 0.39
2 4_clus__clu_score : 0.35
dtype: object
VALUE : 0.4067
--------- Raw Scores ---------
6_post scores:
None
Finished running step!
<cartodata.model_selection.utils.Results object at 0x7efd4413cbb0>
.. GENERATED FROM PYTHON SOURCE LINES 472-473
We can see that we have run the first 6 parameter sets in the dataframe.
.. GENERATED FROM PYTHON SOURCE LINES 473-476
.. code-block:: Python
len(experiment.results.runs_)
.. rst-class:: sphx-glr-script-out
.. code-block:: none
6
.. GENERATED FROM PYTHON SOURCE LINES 477-478
Let's say we have stopped the experiment at this point. The parameter list that we have run already is saved as a .CSV file in the `TOP_DIR`.
.. GENERATED FROM PYTHON SOURCE LINES 478-488
.. code-block:: Python
# !ls $TOP_DIR
""
import pandas as pd
df = pd.read_csv(TOP_DIR / "experiment.csv", index_col=0)
df
.. raw:: html
<div class="output_subarea output_html rendered_html output_result">
<div>
<style scoped>
.dataframe tbody tr th:only-of-type {
vertical-align: middle;
}
.dataframe tbody tr th {
vertical-align: top;
}
.dataframe thead th {
text-align: right;
}
</style>
<table border="1" class="dataframe">
<thead>
<tr style="text-align: right;">
<th></th>
<th>id</th>
<th>robustseed</th>
<th>authors__filter_min_score</th>
<th>filter_min_score</th>
<th>projection_nD</th>
<th>projection_2D</th>
<th>selected</th>
</tr>
</thead>
<tbody>
<tr>
<th>0</th>
<td>68d289ffd85ae917e710</td>
<td>0</td>
<td>4</td>
<td>6</td>
<td>{'key': 'lsa', 'num_dims': 50, 'extra_param': ...</td>
<td>{'key': 'umap', 'n_neighbors': 10, 'min_dist':...</td>
<td>True</td>
</tr>
<tr>
<th>1</th>
<td>ce1c0d334fc75a6570a5</td>
<td>0</td>
<td>4</td>
<td>6</td>
<td>{'key': 'lsa', 'num_dims': 50, 'extra_param': ...</td>
<td>{'key': 'umap', 'n_neighbors': 10, 'min_dist':...</td>
<td>True</td>
</tr>
<tr>
<th>2</th>
<td>a588bce09630b37c4111</td>
<td>0</td>
<td>4</td>
<td>6</td>
<td>{'key': 'lsa', 'num_dims': 50, 'extra_param': ...</td>
<td>{'key': 'umap', 'n_neighbors': 10, 'min_dist':...</td>
<td>True</td>
</tr>
<tr>
<th>3</th>
<td>06128c38fd13123e263c</td>
<td>0</td>
<td>4</td>
<td>6</td>
<td>{'key': 'lsa', 'num_dims': 50, 'extra_param': ...</td>
<td>{'key': 'umap', 'n_neighbors': 20, 'min_dist':...</td>
<td>True</td>
</tr>
<tr>
<th>4</th>
<td>1040be9e44c8ab04aa0e</td>
<td>0</td>
<td>4</td>
<td>6</td>
<td>{'key': 'lsa', 'num_dims': 50, 'extra_param': ...</td>
<td>{'key': 'umap', 'n_neighbors': 20, 'min_dist':...</td>
<td>True</td>
</tr>
<tr>
<th>5</th>
<td>014dbb63a341d2622f25</td>
<td>0</td>
<td>4</td>
<td>6</td>
<td>{'key': 'lsa', 'num_dims': 50, 'extra_param': ...</td>
<td>{'key': 'umap', 'n_neighbors': 20, 'min_dist':...</td>
<td>True</td>
</tr>
<tr>
<th>6</th>
<td>06a7ca0915d11a6cad00</td>
<td>0</td>
<td>4</td>
<td>6</td>
<td>{'key': 'lsa', 'num_dims': 50, 'extra_param': ...</td>
<td>{'key': 'umap', 'n_neighbors': 50, 'min_dist':...</td>
<td>False</td>
</tr>
<tr>
<th>7</th>
<td>979bdf3fc81db6766d24</td>
<td>0</td>
<td>4</td>
<td>6</td>
<td>{'key': 'lsa', 'num_dims': 50, 'extra_param': ...</td>
<td>{'key': 'umap', 'n_neighbors': 50, 'min_dist':...</td>
<td>False</td>
</tr>
<tr>
<th>8</th>
<td>74f9b611573d48c19a8e</td>
<td>0</td>
<td>4</td>
<td>6</td>
<td>{'key': 'lsa', 'num_dims': 50, 'extra_param': ...</td>
<td>{'key': 'umap', 'n_neighbors': 50, 'min_dist':...</td>
<td>False</td>
</tr>
<tr>
<th>9</th>
<td>c5ec62b56a0c0563c079</td>
<td>0</td>
<td>4</td>
<td>6</td>
<td>{'key': 'lsa', 'num_dims': 50, 'extra_param': ...</td>
<td>{'key': 'umap', 'n_neighbors': 10, 'min_dist':...</td>
<td>False</td>
</tr>
<tr>
<th>10</th>
<td>b11b71ed11b157e13f95</td>
<td>0</td>
<td>4</td>
<td>6</td>
<td>{'key': 'lsa', 'num_dims': 50, 'extra_param': ...</td>
<td>{'key': 'umap', 'n_neighbors': 10, 'min_dist':...</td>
<td>False</td>
</tr>
<tr>
<th>11</th>
<td>a7909c645cf88ed5c209</td>
<td>0</td>
<td>4</td>
<td>6</td>
<td>{'key': 'lsa', 'num_dims': 50, 'extra_param': ...</td>
<td>{'key': 'umap', 'n_neighbors': 10, 'min_dist':...</td>
<td>False</td>
</tr>
<tr>
<th>12</th>
<td>4c16fefcccebf0b96309</td>
<td>0</td>
<td>4</td>
<td>6</td>
<td>{'key': 'lsa', 'num_dims': 50, 'extra_param': ...</td>
<td>{'key': 'umap', 'n_neighbors': 20, 'min_dist':...</td>
<td>False</td>
</tr>
<tr>
<th>13</th>
<td>97a0b8354402e18fa744</td>
<td>0</td>
<td>4</td>
<td>6</td>
<td>{'key': 'lsa', 'num_dims': 50, 'extra_param': ...</td>
<td>{'key': 'umap', 'n_neighbors': 20, 'min_dist':...</td>
<td>False</td>
</tr>
<tr>
<th>14</th>
<td>dcfa478890363f0396c3</td>
<td>0</td>
<td>4</td>
<td>6</td>
<td>{'key': 'lsa', 'num_dims': 50, 'extra_param': ...</td>
<td>{'key': 'umap', 'n_neighbors': 20, 'min_dist':...</td>
<td>False</td>
</tr>
<tr>
<th>15</th>
<td>0070d0c653c88c823b5c</td>
<td>0</td>
<td>4</td>
<td>6</td>
<td>{'key': 'lsa', 'num_dims': 50, 'extra_param': ...</td>
<td>{'key': 'umap', 'n_neighbors': 50, 'min_dist':...</td>
<td>False</td>
</tr>
<tr>
<th>16</th>
<td>9fd981e9c09b2eed1de6</td>
<td>0</td>
<td>4</td>
<td>6</td>
<td>{'key': 'lsa', 'num_dims': 50, 'extra_param': ...</td>
<td>{'key': 'umap', 'n_neighbors': 50, 'min_dist':...</td>
<td>False</td>
</tr>
<tr>
<th>17</th>
<td>8ff3e42500d4389f445d</td>
<td>0</td>
<td>4</td>
<td>6</td>
<td>{'key': 'lsa', 'num_dims': 50, 'extra_param': ...</td>
<td>{'key': 'umap', 'n_neighbors': 50, 'min_dist':...</td>
<td>False</td>
</tr>
<tr>
<th>18</th>
<td>9bd0b3935ca3e4457d8a</td>
<td>0</td>
<td>4</td>
<td>6</td>
<td>{'key': 'lsa', 'num_dims': 100, 'extra_param':...</td>
<td>{'key': 'umap', 'n_neighbors': 10, 'min_dist':...</td>
<td>False</td>
</tr>
<tr>
<th>19</th>
<td>9a6fd2af303ac5430413</td>
<td>0</td>
<td>4</td>
<td>6</td>
<td>{'key': 'lsa', 'num_dims': 100, 'extra_param':...</td>
<td>{'key': 'umap', 'n_neighbors': 10, 'min_dist':...</td>
<td>False</td>
</tr>
<tr>
<th>20</th>
<td>151ba7fff2224cb7d57b</td>
<td>0</td>
<td>4</td>
<td>6</td>
<td>{'key': 'lsa', 'num_dims': 100, 'extra_param':...</td>
<td>{'key': 'umap', 'n_neighbors': 10, 'min_dist':...</td>
<td>False</td>
</tr>
<tr>
<th>21</th>
<td>45e2d7c0f5ed87a9653b</td>
<td>0</td>
<td>4</td>
<td>6</td>
<td>{'key': 'lsa', 'num_dims': 100, 'extra_param':...</td>
<td>{'key': 'umap', 'n_neighbors': 20, 'min_dist':...</td>
<td>False</td>
</tr>
<tr>
<th>22</th>
<td>fb97932fdc7a24071f48</td>
<td>0</td>
<td>4</td>
<td>6</td>
<td>{'key': 'lsa', 'num_dims': 100, 'extra_param':...</td>
<td>{'key': 'umap', 'n_neighbors': 20, 'min_dist':...</td>
<td>False</td>
</tr>
<tr>
<th>23</th>
<td>3381a66259bc4e6d8899</td>
<td>0</td>
<td>4</td>
<td>6</td>
<td>{'key': 'lsa', 'num_dims': 100, 'extra_param':...</td>
<td>{'key': 'umap', 'n_neighbors': 20, 'min_dist':...</td>
<td>False</td>
</tr>
<tr>
<th>24</th>
<td>5c605c552b449a76cbbf</td>
<td>0</td>
<td>4</td>
<td>6</td>
<td>{'key': 'lsa', 'num_dims': 100, 'extra_param':...</td>
<td>{'key': 'umap', 'n_neighbors': 50, 'min_dist':...</td>
<td>False</td>
</tr>
<tr>
<th>25</th>
<td>eae12b9b57e5abbf5b3b</td>
<td>0</td>
<td>4</td>
<td>6</td>
<td>{'key': 'lsa', 'num_dims': 100, 'extra_param':...</td>
<td>{'key': 'umap', 'n_neighbors': 50, 'min_dist':...</td>
<td>False</td>
</tr>
<tr>
<th>26</th>
<td>1595a61c3d93a94b121e</td>
<td>0</td>
<td>4</td>
<td>6</td>
<td>{'key': 'lsa', 'num_dims': 100, 'extra_param':...</td>
<td>{'key': 'umap', 'n_neighbors': 50, 'min_dist':...</td>
<td>False</td>
</tr>
<tr>
<th>27</th>
<td>d1ced9d06c4ed06937ad</td>
<td>0</td>
<td>4</td>
<td>6</td>
<td>{'key': 'lsa', 'num_dims': 100, 'extra_param':...</td>
<td>{'key': 'umap', 'n_neighbors': 10, 'min_dist':...</td>
<td>False</td>
</tr>
<tr>
<th>28</th>
<td>fc5671119a7e46e9b519</td>
<td>0</td>
<td>4</td>
<td>6</td>
<td>{'key': 'lsa', 'num_dims': 100, 'extra_param':...</td>
<td>{'key': 'umap', 'n_neighbors': 10, 'min_dist':...</td>
<td>False</td>
</tr>
<tr>
<th>29</th>
<td>c6c7cfaf3dab327815a3</td>
<td>0</td>
<td>4</td>
<td>6</td>
<td>{'key': 'lsa', 'num_dims': 100, 'extra_param':...</td>
<td>{'key': 'umap', 'n_neighbors': 10, 'min_dist':...</td>
<td>False</td>
</tr>
<tr>
<th>30</th>
<td>e3c9335d86928f04d986</td>
<td>0</td>
<td>4</td>
<td>6</td>
<td>{'key': 'lsa', 'num_dims': 100, 'extra_param':...</td>
<td>{'key': 'umap', 'n_neighbors': 20, 'min_dist':...</td>
<td>False</td>
</tr>
<tr>
<th>31</th>
<td>b2b14d84c6cf741b3f94</td>
<td>0</td>
<td>4</td>
<td>6</td>
<td>{'key': 'lsa', 'num_dims': 100, 'extra_param':...</td>
<td>{'key': 'umap', 'n_neighbors': 20, 'min_dist':...</td>
<td>False</td>
</tr>
<tr>
<th>32</th>
<td>e079875025cff951a37d</td>
<td>0</td>
<td>4</td>
<td>6</td>
<td>{'key': 'lsa', 'num_dims': 100, 'extra_param':...</td>
<td>{'key': 'umap', 'n_neighbors': 20, 'min_dist':...</td>
<td>False</td>
</tr>
<tr>
<th>33</th>
<td>943a271c14a11b4e5954</td>
<td>0</td>
<td>4</td>
<td>6</td>
<td>{'key': 'lsa', 'num_dims': 100, 'extra_param':...</td>
<td>{'key': 'umap', 'n_neighbors': 50, 'min_dist':...</td>
<td>False</td>
</tr>
<tr>
<th>34</th>
<td>fea341b317d4d596ea82</td>
<td>0</td>
<td>4</td>
<td>6</td>
<td>{'key': 'lsa', 'num_dims': 100, 'extra_param':...</td>
<td>{'key': 'umap', 'n_neighbors': 50, 'min_dist':...</td>
<td>False</td>
</tr>
<tr>
<th>35</th>
<td>ba4951c2c78e0031fb8e</td>
<td>0</td>
<td>4</td>
<td>6</td>
<td>{'key': 'lsa', 'num_dims': 100, 'extra_param':...</td>
<td>{'key': 'umap', 'n_neighbors': 50, 'min_dist':...</td>
<td>False</td>
</tr>
<tr>
<th>36</th>
<td>939f3d80d6ec35db69f7</td>
<td>0</td>
<td>4</td>
<td>6</td>
<td>{'key': 'bert', 'family': 'all-MiniLM-L6-v2'}</td>
<td>{'key': 'umap', 'n_neighbors': 10, 'min_dist':...</td>
<td>False</td>
</tr>
<tr>
<th>37</th>
<td>18a385a32e9a8250cfda</td>
<td>0</td>
<td>4</td>
<td>6</td>
<td>{'key': 'bert', 'family': 'all-MiniLM-L6-v2'}</td>
<td>{'key': 'umap', 'n_neighbors': 10, 'min_dist':...</td>
<td>False</td>
</tr>
<tr>
<th>38</th>
<td>41f146808f2ef8b013be</td>
<td>0</td>
<td>4</td>
<td>6</td>
<td>{'key': 'bert', 'family': 'all-MiniLM-L6-v2'}</td>
<td>{'key': 'umap', 'n_neighbors': 10, 'min_dist':...</td>
<td>False</td>
</tr>
<tr>
<th>39</th>
<td>f32ac02bc42412221a98</td>
<td>0</td>
<td>4</td>
<td>6</td>
<td>{'key': 'bert', 'family': 'all-MiniLM-L6-v2'}</td>
<td>{'key': 'umap', 'n_neighbors': 20, 'min_dist':...</td>
<td>False</td>
</tr>
<tr>
<th>40</th>
<td>ed7b6ebc7f1b57baa10c</td>
<td>0</td>
<td>4</td>
<td>6</td>
<td>{'key': 'bert', 'family': 'all-MiniLM-L6-v2'}</td>
<td>{'key': 'umap', 'n_neighbors': 20, 'min_dist':...</td>
<td>False</td>
</tr>
<tr>
<th>41</th>
<td>9616f8655120a922eb8f</td>
<td>0</td>
<td>4</td>
<td>6</td>
<td>{'key': 'bert', 'family': 'all-MiniLM-L6-v2'}</td>
<td>{'key': 'umap', 'n_neighbors': 20, 'min_dist':...</td>
<td>False</td>
</tr>
<tr>
<th>42</th>
<td>a72244820858b67def57</td>
<td>0</td>
<td>4</td>
<td>6</td>
<td>{'key': 'bert', 'family': 'all-MiniLM-L6-v2'}</td>
<td>{'key': 'umap', 'n_neighbors': 50, 'min_dist':...</td>
<td>False</td>
</tr>
<tr>
<th>43</th>
<td>3ae9168b3636e9c9ab4f</td>
<td>0</td>
<td>4</td>
<td>6</td>
<td>{'key': 'bert', 'family': 'all-MiniLM-L6-v2'}</td>
<td>{'key': 'umap', 'n_neighbors': 50, 'min_dist':...</td>
<td>False</td>
</tr>
<tr>
<th>44</th>
<td>ec8c846300e8221b9a52</td>
<td>0</td>
<td>4</td>
<td>6</td>
<td>{'key': 'bert', 'family': 'all-MiniLM-L6-v2'}</td>
<td>{'key': 'umap', 'n_neighbors': 50, 'min_dist':...</td>
<td>False</td>
</tr>
</tbody>
</table>
</div>
</div>
<br />
<br />
.. GENERATED FROM PYTHON SOURCE LINES 489-490
When we want to continue to run the experiment, we initiate a new parameter iterator with this file and continue running the experiment for the parameters set that were not used during the previous experiments.
.. GENERATED FROM PYTHON SOURCE LINES 490-521
.. code-block:: Python
experiment_file = TOP_DIR / "experiment.csv"
param_iterator = GridIterator(csv_filepath=experiment_file)
""
experiment = Experiment(
"lisn", "2022.11.15.1", TOP_DIR, CONF_DIR, INPUT_DIR,
NATURE, SOURCE, param_iterator,
score_list=[CustomScore,
NeighborsND,
Neighbors2D,
Comparative,
# TrustworthinessSklearn,
# TrustworthinessUmap,
Clustering,
FinalScore],
final_score__name_list=[
PhaseProjectionND.prefix("neighbors_articles_authors"),
PhaseProjection2D.prefix("neighbors_articles_authors"),
PhaseClustering.prefix("clu_score")],
neighbors__recompute=True,
neighbors__min_score =30,
# trustworthiness_sklearn__n_neighbors=10,
custom__factor=10
)
""
experiment.run(2)
.. rst-class:: sphx-glr-script-out
.. code-block:: none
Running experiment for parameters:
{'id': '06a7ca0915d11a6cad00', 'robustseed': 0, 'authors__filter_min_score': 4, 'filter_min_score': 6, 'projection_nD': {'key': 'lsa', 'num_dims': 50, 'extra_param': True}, 'projection_2D': {'key': 'umap', 'n_neighbors': 50, 'min_dist': 0.1, 'metric': 'euclidean'}}
Matrices of type mat for articles, authors, teams, labs, words already exist. Skipping creation.
Matrices of type lsa for articles, authors, teams, labs, words already exist. Skipping creation.
================================
Run params : {'2_nD__rkey': 'lsa', '2_nD__rnum_dims': 50, '2_nD__rnormalize': True, '2_nD__rextra_param': True}
--------------------------------
Scoring params : {'2_nD__smin_score': 30, '2_nD__srecompute': True, '2_nD__ssample_size': None, '2_nD__sn_neighbors': 10, '2_nD__srandom_state': 42}
================================
----------- Scores -----------
2_nD__neighbors_articles_authors : 0.4779
--------- Desc Scores ---------
2_nD__neighbors_articles_authors_det
0 Céline Gicquel : 0.82
1 Loïc Paulevé : 0.73
2 Nikolaus Hansen : 0.69
3 Sébastien Tixeuil : 0.68
4 Philippe Caillou : 0.67
5 Cyril Furtlehner : 0.67
6 Raymond Ros : 0.66
7 Isabelle Guyon : 0.62
8 Dimo Brockhoff : 0.61
9 Paola Tubaro : 0.60
10 Anne Auger : 0.59
11 Yann Ponty : 0.59
12 Sarah Cohen-Boulakia : 0.58
13 Olivier Teytaud : 0.57
14 Nicolas Bredeche : 0.54
15 Marc Baboulin : 0.54
16 Tobias Isenberg : 0.54
17 Albert Cohen : 0.53
18 Guillaume Charpiat : 0.52
19 Chantal Reynaud : 0.51
20 Fatiha Saïs : 0.51
21 Ioana Manolescu : 0.51
22 Sylvie Boldo : 0.49
23 Lonni Besançon : 0.49
24 Nathalie Pernelle : 0.48
25 Franck Cappello : 0.48
26 Petra Isenberg : 0.48
27 Guillaume Melquiond : 0.47
28 Claude Marché : 0.47
29 Marc Schoenauer : 0.44
30 Jean-Daniel Fekete : 0.44
31 François Goasdoué : 0.43
32 Fatiha Zaidi : 0.39
33 Olivier Chapuis : 0.38
34 Michèle Sebag : 0.38
35 Nathann Cohen : 0.38
36 Steven Martin : 0.36
37 Anastasia Bezerianos : 0.32
38 Pierre Dragicevic : 0.32
39 Wendy Mackay : 0.30
40 Michel Beaudouin-Lafon : 0.30
41 Alain Denise : 0.30
42 Balázs Kégl : 0.29
43 Caroline Appert : 0.27
44 Evelyne Lutton : 0.27
45 Johanne Cohen : 0.27
46 Emmanuel Pietriga : 0.26
47 Wendy E. Mackay : 0.20
Name: 0, dtype: object
VALUE : 0.4779
--------- Raw Scores ---------
2_nD__neighbors_articles_authors
Sébastien Tixeuil 0.680851
Michèle Sebag 0.383942
Johanne Cohen 0.272093
Albert Cohen 0.529310
Wendy E. Mackay 0.200000
Philippe Caillou 0.674419
Alain Denise 0.297222
Jean-Daniel Fekete 0.437107
Emmanuel Pietriga 0.260317
Yann Ponty 0.586364
Marc Schoenauer 0.443165
Franck Cappello 0.480488
Caroline Appert 0.273913
Michel Beaudouin-Lafon 0.297531
Wendy Mackay 0.302128
Anne Auger 0.591139
Evelyne Lutton 0.272973
Pierre Dragicevic 0.317284
Ioana Manolescu 0.507317
Nikolaus Hansen 0.686420
Nicolas Bredeche 0.544118
Olivier Teytaud 0.567925
François Goasdoué 0.430189
Nathalie Pernelle 0.482353
Fatiha Saïs 0.507317
Sarah Cohen-Boulakia 0.581818
Claude Marché 0.465957
Chantal Reynaud 0.513333
Olivier Chapuis 0.384615
Steven Martin 0.356410
Fatiha Zaidi 0.390625
Balázs Kégl 0.286842
Paola Tubaro 0.602564
Raymond Ros 0.655882
Cyril Furtlehner 0.666667
Anastasia Bezerianos 0.319403
Sylvie Boldo 0.488571
Guillaume Melquiond 0.472727
Marc Baboulin 0.540000
Dimo Brockhoff 0.610256
Nathann Cohen 0.380488
Petra Isenberg 0.479439
Tobias Isenberg 0.536752
Loïc Paulevé 0.726190
Céline Gicquel 0.818421
Isabelle Guyon 0.623596
Guillaume Charpiat 0.522581
Lonni Besançon 0.487879
Name: 0, dtype: float64
================================
Run params : {}
--------------------------------
Scoring params : {}
================================
----------- Scores -----------
custom_score_lsa : 10.0000
--------- Desc Scores ---------
--------- Raw Scores ---------
custom scores:
None
================================
Run params : {'2_nD__rkey': 'lsa', '2_nD__rnum_dims': 50, '2_nD__rnormalize': True, '2_nD__rextra_param': True}
--------------------------------
Scoring params : {'2_nD__smin_score': 30, '2_nD__srecompute': True, '2_nD__ssample_size': None, '2_nD__sn_neighbors': 10, '2_nD__srandom_state': 42, '2_nD__sfactor': 10}
================================
----------- Scores -----------
2_nD__neighbors_articles_authors : 0.4779
2_nD__custom_score_lsa : 10.0000
--------- Desc Scores ---------
2_nD__neighbors_articles_authors_det
0 Céline Gicquel : 0.82
1 Loïc Paulevé : 0.73
2 Nikolaus Hansen : 0.69
3 Sébastien Tixeuil : 0.68
4 Philippe Caillou : 0.67
5 Cyril Furtlehner : 0.67
6 Raymond Ros : 0.66
7 Isabelle Guyon : 0.62
8 Dimo Brockhoff : 0.61
9 Paola Tubaro : 0.60
10 Anne Auger : 0.59
11 Yann Ponty : 0.59
12 Sarah Cohen-Boulakia : 0.58
13 Olivier Teytaud : 0.57
14 Nicolas Bredeche : 0.54
15 Marc Baboulin : 0.54
16 Tobias Isenberg : 0.54
17 Albert Cohen : 0.53
18 Guillaume Charpiat : 0.52
19 Chantal Reynaud : 0.51
20 Fatiha Saïs : 0.51
21 Ioana Manolescu : 0.51
22 Sylvie Boldo : 0.49
23 Lonni Besançon : 0.49
24 Nathalie Pernelle : 0.48
25 Franck Cappello : 0.48
26 Petra Isenberg : 0.48
27 Guillaume Melquiond : 0.47
28 Claude Marché : 0.47
29 Marc Schoenauer : 0.44
30 Jean-Daniel Fekete : 0.44
31 François Goasdoué : 0.43
32 Fatiha Zaidi : 0.39
33 Olivier Chapuis : 0.38
34 Michèle Sebag : 0.38
35 Nathann Cohen : 0.38
36 Steven Martin : 0.36
37 Anastasia Bezerianos : 0.32
38 Pierre Dragicevic : 0.32
39 Wendy Mackay : 0.30
40 Michel Beaudouin-Lafon : 0.30
41 Alain Denise : 0.30
42 Balázs Kégl : 0.29
43 Caroline Appert : 0.27
44 Evelyne Lutton : 0.27
45 Johanne Cohen : 0.27
46 Emmanuel Pietriga : 0.26
47 Wendy E. Mackay : 0.20
Name: 0, dtype: object
VALUE : 0.4779
--------- Raw Scores ---------
2_nD__neighbors_articles_authors
Sébastien Tixeuil 0.680851
Michèle Sebag 0.383942
Johanne Cohen 0.272093
Albert Cohen 0.529310
Wendy E. Mackay 0.200000
Philippe Caillou 0.674419
Alain Denise 0.297222
Jean-Daniel Fekete 0.437107
Emmanuel Pietriga 0.260317
Yann Ponty 0.586364
Marc Schoenauer 0.443165
Franck Cappello 0.480488
Caroline Appert 0.273913
Michel Beaudouin-Lafon 0.297531
Wendy Mackay 0.302128
Anne Auger 0.591139
Evelyne Lutton 0.272973
Pierre Dragicevic 0.317284
Ioana Manolescu 0.507317
Nikolaus Hansen 0.686420
Nicolas Bredeche 0.544118
Olivier Teytaud 0.567925
François Goasdoué 0.430189
Nathalie Pernelle 0.482353
Fatiha Saïs 0.507317
Sarah Cohen-Boulakia 0.581818
Claude Marché 0.465957
Chantal Reynaud 0.513333
Olivier Chapuis 0.384615
Steven Martin 0.356410
Fatiha Zaidi 0.390625
Balázs Kégl 0.286842
Paola Tubaro 0.602564
Raymond Ros 0.655882
Cyril Furtlehner 0.666667
Anastasia Bezerianos 0.319403
Sylvie Boldo 0.488571
Guillaume Melquiond 0.472727
Marc Baboulin 0.540000
Dimo Brockhoff 0.610256
Nathann Cohen 0.380488
Petra Isenberg 0.479439
Tobias Isenberg 0.536752
Loïc Paulevé 0.726190
Céline Gicquel 0.818421
Isabelle Guyon 0.623596
Guillaume Charpiat 0.522581
Lonni Besançon 0.487879
Name: 0, dtype: float64
2_projection_nD scores:
None
umap matrices generated.
================================
Run params : {'3_2D__rkey': 'umap', '3_2D__rmetric': 'euclidean', '3_2D__rn_neighbors': 50, '3_2D__rmin_dist': 0.1, '3_2D__rinit': 'random', '3_2D__rlearning_rate': 1.0, '3_2D__rn_epochs': None, '3_2D__rrandom_state': None}
--------------------------------
Scoring params : {'3_2D__smin_score': 30, '3_2D__srecompute': True, '3_2D__ssample_size': None, '3_2D__sn_neighbors': 10, '3_2D__srandom_state': 42}
================================
----------- Scores -----------
3_2D__neighbors_articles_authors : 0.4092
3_2D__2_nD_3_2D_neighbors_articles_authors_dim_mean : -0.0687
3_2D__2_nD_3_2D_neighbors_articles_authors_dim_median : -0.0623
--------- Desc Scores ---------
3_2D__neighbors_articles_authors_det
0 Céline Gicquel : 0.69
1 Loïc Paulevé : 0.67
2 Nikolaus Hansen : 0.60
3 Sébastien Tixeuil : 0.59
4 Cyril Furtlehner : 0.59
5 Isabelle Guyon : 0.56
6 Yann Ponty : 0.55
7 Nicolas Bredeche : 0.55
8 Philippe Caillou : 0.54
9 Raymond Ros : 0.54
10 Anne Auger : 0.52
11 Tobias Isenberg : 0.52
12 Dimo Brockhoff : 0.51
13 Albert Cohen : 0.50
14 Olivier Teytaud : 0.50
15 Fatiha Saïs : 0.48
16 Marc Baboulin : 0.46
17 Paola Tubaro : 0.45
18 Nathalie Pernelle : 0.44
19 Ioana Manolescu : 0.44
20 Sarah Cohen-Boulakia : 0.43
21 Sylvie Boldo : 0.42
22 Petra Isenberg : 0.41
23 Chantal Reynaud : 0.41
24 Marc Schoenauer : 0.40
25 Lonni Besançon : 0.40
26 Jean-Daniel Fekete : 0.40
27 Nathann Cohen : 0.40
28 Franck Cappello : 0.39
29 Claude Marché : 0.35
30 François Goasdoué : 0.33
31 Guillaume Melquiond : 0.33
32 Guillaume Charpiat : 0.33
33 Olivier Chapuis : 0.33
34 Michèle Sebag : 0.33
35 Steven Martin : 0.31
36 Pierre Dragicevic : 0.30
37 Fatiha Zaidi : 0.28
38 Anastasia Bezerianos : 0.28
39 Caroline Appert : 0.26
40 Michel Beaudouin-Lafon : 0.26
41 Alain Denise : 0.25
42 Johanne Cohen : 0.24
43 Balázs Kégl : 0.23
44 Evelyne Lutton : 0.23
45 Wendy Mackay : 0.22
46 Wendy E. Mackay : 0.21
47 Emmanuel Pietriga : 0.20
dtype: object
VALUE : 0.4092
3_2D__2_nD_3_2D_neighbors_articles_authors_dim_det
0 Guillaume Charpiat : -0.19
1 Paola Tubaro : -0.15
2 Sarah Cohen-Boulakia : -0.15
3 Guillaume Melquiond : -0.14
4 Philippe Caillou : -0.13
5 Céline Gicquel : -0.13
6 Claude Marché : -0.12
7 Raymond Ros : -0.11
8 Chantal Reynaud : -0.11
9 Fatiha Zaidi : -0.11
10 Dimo Brockhoff : -0.10
11 François Goasdoué : -0.10
12 Franck Cappello : -0.09
13 Sébastien Tixeuil : -0.09
14 Nikolaus Hansen : -0.09
15 Lonni Besançon : -0.09
16 Marc Baboulin : -0.08
17 Wendy Mackay : -0.08
18 Cyril Furtlehner : -0.08
19 Ioana Manolescu : -0.07
20 Petra Isenberg : -0.07
21 Anne Auger : -0.07
22 Sylvie Boldo : -0.07
23 Olivier Teytaud : -0.07
24 Isabelle Guyon : -0.06
25 Loïc Paulevé : -0.06
26 Michèle Sebag : -0.06
27 Emmanuel Pietriga : -0.06
28 Olivier Chapuis : -0.06
29 Balázs Kégl : -0.05
30 Alain Denise : -0.05
31 Steven Martin : -0.05
32 Nathalie Pernelle : -0.04
33 Marc Schoenauer : -0.04
34 Anastasia Bezerianos : -0.04
35 Michel Beaudouin-Lafon : -0.04
36 Evelyne Lutton : -0.04
37 Jean-Daniel Fekete : -0.04
38 Yann Ponty : -0.04
39 Johanne Cohen : -0.03
40 Fatiha Saïs : -0.03
41 Albert Cohen : -0.03
42 Tobias Isenberg : -0.02
43 Pierre Dragicevic : -0.01
44 Caroline Appert : -0.01
45 Nicolas Bredeche : 0.01
46 Wendy E. Mackay : 0.01
47 Nathann Cohen : 0.01
dtype: object
VALUE : -0.0687
--------- Raw Scores ---------
3_2D__neighbors_articles_authors
Sébastien Tixeuil 0.591489
Michèle Sebag 0.325547
Johanne Cohen 0.239535
Albert Cohen 0.503448
Wendy E. Mackay 0.208696
Philippe Caillou 0.544186
Alain Denise 0.250000
Jean-Daniel Fekete 0.399371
Emmanuel Pietriga 0.203175
Yann Ponty 0.550000
Marc Schoenauer 0.400719
Franck Cappello 0.387805
Caroline Appert 0.263043
Michel Beaudouin-Lafon 0.256790
Wendy Mackay 0.221277
Anne Auger 0.522785
Evelyne Lutton 0.232432
Pierre Dragicevic 0.302469
Ioana Manolescu 0.435366
Nikolaus Hansen 0.597531
Nicolas Bredeche 0.550000
Olivier Teytaud 0.502830
François Goasdoué 0.333962
Nathalie Pernelle 0.438235
Fatiha Saïs 0.475610
Sarah Cohen-Boulakia 0.433333
Claude Marché 0.348936
Chantal Reynaud 0.406667
Olivier Chapuis 0.328846
Steven Martin 0.310256
Fatiha Zaidi 0.284375
Balázs Kégl 0.234211
Paola Tubaro 0.453846
Raymond Ros 0.544118
Cyril Furtlehner 0.589744
Anastasia Bezerianos 0.277612
Sylvie Boldo 0.422857
Guillaume Melquiond 0.330303
Marc Baboulin 0.457778
Dimo Brockhoff 0.510256
Nathann Cohen 0.395122
Petra Isenberg 0.409346
Tobias Isenberg 0.516239
Loïc Paulevé 0.666667
Céline Gicquel 0.689474
Isabelle Guyon 0.564045
Guillaume Charpiat 0.329032
Lonni Besançon 0.400000
dtype: float64
3_2D__2_nD_3_2D_neighbors_articles_authors_dim
2_nD__neighbors_articles_authors \
Sébastien Tixeuil 0.680851
Michèle Sebag 0.383942
Johanne Cohen 0.272093
Albert Cohen 0.529310
Wendy E. Mackay 0.200000
Philippe Caillou 0.674419
Alain Denise 0.297222
Jean-Daniel Fekete 0.437107
Emmanuel Pietriga 0.260317
Yann Ponty 0.586364
Marc Schoenauer 0.443165
Franck Cappello 0.480488
Caroline Appert 0.273913
Michel Beaudouin-Lafon 0.297531
Wendy Mackay 0.302128
Anne Auger 0.591139
Evelyne Lutton 0.272973
Pierre Dragicevic 0.317284
Ioana Manolescu 0.507317
Nikolaus Hansen 0.686420
Nicolas Bredeche 0.544118
Olivier Teytaud 0.567925
François Goasdoué 0.430189
Nathalie Pernelle 0.482353
Fatiha Saïs 0.507317
Sarah Cohen-Boulakia 0.581818
Claude Marché 0.465957
Chantal Reynaud 0.513333
Olivier Chapuis 0.384615
Steven Martin 0.356410
Fatiha Zaidi 0.390625
Balázs Kégl 0.286842
Paola Tubaro 0.602564
Raymond Ros 0.655882
Cyril Furtlehner 0.666667
Anastasia Bezerianos 0.319403
Sylvie Boldo 0.488571
Guillaume Melquiond 0.472727
Marc Baboulin 0.540000
Dimo Brockhoff 0.610256
Nathann Cohen 0.380488
Petra Isenberg 0.479439
Tobias Isenberg 0.536752
Loïc Paulevé 0.726190
Céline Gicquel 0.818421
Isabelle Guyon 0.623596
Guillaume Charpiat 0.522581
Lonni Besançon 0.487879
3_2D__neighbors_articles_authors scdec
Sébastien Tixeuil 0.591489 -0.089362
Michèle Sebag 0.325547 -0.058394
Johanne Cohen 0.239535 -0.032558
Albert Cohen 0.503448 -0.025862
Wendy E. Mackay 0.208696 0.008696
Philippe Caillou 0.544186 -0.130233
Alain Denise 0.250000 -0.047222
Jean-Daniel Fekete 0.399371 -0.037736
Emmanuel Pietriga 0.203175 -0.057143
Yann Ponty 0.550000 -0.036364
Marc Schoenauer 0.400719 -0.042446
Franck Cappello 0.387805 -0.092683
Caroline Appert 0.263043 -0.010870
Michel Beaudouin-Lafon 0.256790 -0.040741
Wendy Mackay 0.221277 -0.080851
Anne Auger 0.522785 -0.068354
Evelyne Lutton 0.232432 -0.040541
Pierre Dragicevic 0.302469 -0.014815
Ioana Manolescu 0.435366 -0.071951
Nikolaus Hansen 0.597531 -0.088889
Nicolas Bredeche 0.550000 0.005882
Olivier Teytaud 0.502830 -0.065094
François Goasdoué 0.333962 -0.096226
Nathalie Pernelle 0.438235 -0.044118
Fatiha Saïs 0.475610 -0.031707
Sarah Cohen-Boulakia 0.433333 -0.148485
Claude Marché 0.348936 -0.117021
Chantal Reynaud 0.406667 -0.106667
Olivier Chapuis 0.328846 -0.055769
Steven Martin 0.310256 -0.046154
Fatiha Zaidi 0.284375 -0.106250
Balázs Kégl 0.234211 -0.052632
Paola Tubaro 0.453846 -0.148718
Raymond Ros 0.544118 -0.111765
Cyril Furtlehner 0.589744 -0.076923
Anastasia Bezerianos 0.277612 -0.041791
Sylvie Boldo 0.422857 -0.065714
Guillaume Melquiond 0.330303 -0.142424
Marc Baboulin 0.457778 -0.082222
Dimo Brockhoff 0.510256 -0.100000
Nathann Cohen 0.395122 0.014634
Petra Isenberg 0.409346 -0.070093
Tobias Isenberg 0.516239 -0.020513
Loïc Paulevé 0.666667 -0.059524
Céline Gicquel 0.689474 -0.128947
Isabelle Guyon 0.564045 -0.059551
Guillaume Charpiat 0.329032 -0.193548
Lonni Besançon 0.400000 -0.087879
3_projection_2D scores:
None
Nothing in cache, initial Fitting with min_cluster_size=15 Found 75 clusters in 0.2549978460010607s
Max Fitting with min_cluster_size=30 Found 55 clusters in 0.10591133300113142s
Max Fitting with min_cluster_size=60 Found 24 clusters in 0.10550347100070212s
Max Fitting with min_cluster_size=120 Found 2 clusters in 0.1063100969986408s
Midpoint Fitting with min_cluster_size=90 Found 17 clusters in 0.10281582999959937s
Midpoint Fitting with min_cluster_size=105 Found 13 clusters in 0.10239299800014123s
Midpoint Fitting with min_cluster_size=112 Found 13 clusters in 0.10005356100009521s
No need Re-Fitting with min_cluster_size=112
Clusters cached: [2, 13, 13, 17, 24, 55, 75]
Nothing in cache, initial Fitting with min_cluster_size=15 Found 75 clusters in 0.10446891900210176s
Max Fitting with min_cluster_size=30 Found 55 clusters in 0.10560662799980491s
Max Fitting with min_cluster_size=60 Found 24 clusters in 0.10713341199880233s
No need Re-Fitting with min_cluster_size=60
Clusters cached: [24, 55, 75]
================================
Run params : {'4_clus__rkey': 'hdbscan', '4_clus__rbase_factor': 3}
--------------------------------
Scoring params : {'4_clus__siter_stab': 2, '4_clus__sremove_stab': [0, 0.01, 0.03, 0.1, 0.25], '4_clus__smetric': 'euclidean', '4_clus__srandom_state': None}
================================
----------- Scores -----------
4_clus__nb_clust_0 : 8.0000
4_clus__silhouette_0 : 0.1407
4_clus__avg_word_couv_0 : 0.4930
4_clus__med_word_couv_0 : 0.5139
4_clus__avg_word_couv_minus_0 : 0.4675
4_clus__big_small_ratio_0 : 8.3929
4_clus__stab_clus_0 : 0.1385
4_clus__nb_clust_1 : 24.0000
4_clus__silhouette_1 : 0.2960
4_clus__avg_word_couv_1 : 0.6031
4_clus__med_word_couv_1 : 0.5918
4_clus__avg_word_couv_minus_1 : 0.5767
4_clus__big_small_ratio_1 : 10.7377
4_clus__stab_clus_1 : 0.0250
4_clus__avg_stab_avg : 0.0817
4_clus__avg_couv_avg : 0.5481
4_clus__clu_score : 0.3149
--------- Desc Scores ---------
4_clus__clus_eval_pos_0_det
0 query, ontology : s 633 stb 0.10 + 0.32 - 0.03
1 visualizations, interfaces : s 632 stb 0.80 + ...
2 networking internet architecture, neural netwo...
3 optimization control, operations : s 254 stb 0...
4 logic science, semantics : s 252 stb 0.10 + 0....
5 floating point, quantum : s 177 stb 0.30 + 0.4...
6 fluid, numerical simulations : s 145 stb 0.00 ...
7 cluster computing, mobile robots : s 144 stb 0...
8 vertices, minimum : s 135 stb 0.20 + 0.67 - 0.02
9 humanities social sciences, social networks : ...
10 compiler, architectures : s 131 stb 0.00 + 0.6...
11 biology, genes : s 121 stb 0.00 + 0.54 - 0.03
12 covariance matrix adaptation, benchmarking : s...
dtype: object
VALUE : 0.4930
4_clus__clus_eval_pos_1_det
0 networking internet, energy : s 453 stb 0.00 +...
1 touch, virtual : s 331 stb 0.00 + 0.35 - 0.02
2 stochastic, population : s 254 stb 0.00 + 0.41...
3 verification, testing : s 252 stb 0.00 + 0.61 ...
4 information visualization, cognitive science :...
5 floating point arithmetic, lattice : s 177 stb...
6 ontologies, entity : s 171 stb 0.00 + 0.51 - 0.01
7 natural language, translation : s 145 stb 0.00...
8 dynamics, mechanics : s 145 stb 0.00 + 0.57 - ...
9 fault, mobile : s 144 stb 0.00 + 0.62 - 0.04
10 discrete mathematics, vertex : s 135 stb 0.00 ...
11 social sciences, community : s 133 stb 0.00 + ...
12 programming languages, hardware : s 131 stb 0....
13 biological, metabolic : s 121 stb 0.00 + 0.48 ...
14 covariance matrix adaptation evolution, testbe...
15 secondary structure, sequence : s 110 stb 0.00...
16 automata, regulatory : s 91 stb 0.20 + 0.77 - ...
17 documents, materialized views : s 84 stb 0.00 ...
18 services, cloud : s 83 stb 0.00 + 0.90 - 0.03
19 queries, query answers : s 81 stb 0.00 + 0.73 ...
20 monte carlo, games : s 74 stb 0.40 + 0.97 - 0.02
21 linear systems, matrices : s 66 stb 0.00 + 0.7...
22 image : s 62 stb 0.00 + 0.85 - 0.02
23 visual analytics : s 61 stb 0.00 + 0.85 - 0.01
dtype: object
VALUE : 0.6031
--------- Raw Scores ---------
4_clustering scores:
None
['2_nD__neighbors_articles_authors', '3_2D__neighbors_articles_authors', '4_clus__clu_score']
================================
Run params : {}
--------------------------------
Scoring params : {'6_pst__sname_list': ['2_nD__neighbors_articles_authors', '3_2D__neighbors_articles_authors', '4_clus__clu_score']}
================================
----------- Scores -----------
6_pst__final_score : 0.4006
--------- Desc Scores ---------
6_pst__final_score_det
0 2_nD__neighbors_articles_authors : 0.48
1 3_2D__neighbors_articles_authors : 0.41
2 4_clus__clu_score : 0.31
dtype: object
VALUE : 0.4006
--------- Raw Scores ---------
6_post scores:
None
Finished running step!
Running experiment for parameters:
{'id': '979bdf3fc81db6766d24', 'robustseed': 0, 'authors__filter_min_score': 4, 'filter_min_score': 6, 'projection_nD': {'key': 'lsa', 'num_dims': 50, 'extra_param': True}, 'projection_2D': {'key': 'umap', 'n_neighbors': 50, 'min_dist': 0.25, 'metric': 'euclidean'}}
Matrices of type mat for articles, authors, teams, labs, words already exist. Skipping creation.
Matrices of type lsa for articles, authors, teams, labs, words already exist. Skipping creation.
================================
Run params : {'2_nD__rkey': 'lsa', '2_nD__rnum_dims': 50, '2_nD__rnormalize': True, '2_nD__rextra_param': True}
--------------------------------
Scoring params : {'2_nD__smin_score': 30, '2_nD__srecompute': True, '2_nD__ssample_size': None, '2_nD__sn_neighbors': 10, '2_nD__srandom_state': 42}
================================
----------- Scores -----------
2_nD__neighbors_articles_authors : 0.4779
--------- Desc Scores ---------
2_nD__neighbors_articles_authors_det
0 Céline Gicquel : 0.82
1 Loïc Paulevé : 0.73
2 Nikolaus Hansen : 0.69
3 Sébastien Tixeuil : 0.68
4 Philippe Caillou : 0.67
5 Cyril Furtlehner : 0.67
6 Raymond Ros : 0.66
7 Isabelle Guyon : 0.62
8 Dimo Brockhoff : 0.61
9 Paola Tubaro : 0.60
10 Anne Auger : 0.59
11 Yann Ponty : 0.59
12 Sarah Cohen-Boulakia : 0.58
13 Olivier Teytaud : 0.57
14 Nicolas Bredeche : 0.54
15 Marc Baboulin : 0.54
16 Tobias Isenberg : 0.54
17 Albert Cohen : 0.53
18 Guillaume Charpiat : 0.52
19 Chantal Reynaud : 0.51
20 Fatiha Saïs : 0.51
21 Ioana Manolescu : 0.51
22 Sylvie Boldo : 0.49
23 Lonni Besançon : 0.49
24 Nathalie Pernelle : 0.48
25 Franck Cappello : 0.48
26 Petra Isenberg : 0.48
27 Guillaume Melquiond : 0.47
28 Claude Marché : 0.47
29 Marc Schoenauer : 0.44
30 Jean-Daniel Fekete : 0.44
31 François Goasdoué : 0.43
32 Fatiha Zaidi : 0.39
33 Olivier Chapuis : 0.38
34 Michèle Sebag : 0.38
35 Nathann Cohen : 0.38
36 Steven Martin : 0.36
37 Anastasia Bezerianos : 0.32
38 Pierre Dragicevic : 0.32
39 Wendy Mackay : 0.30
40 Michel Beaudouin-Lafon : 0.30
41 Alain Denise : 0.30
42 Balázs Kégl : 0.29
43 Caroline Appert : 0.27
44 Evelyne Lutton : 0.27
45 Johanne Cohen : 0.27
46 Emmanuel Pietriga : 0.26
47 Wendy E. Mackay : 0.20
Name: 0, dtype: object
VALUE : 0.4779
--------- Raw Scores ---------
2_nD__neighbors_articles_authors
Sébastien Tixeuil 0.680851
Michèle Sebag 0.383942
Johanne Cohen 0.272093
Albert Cohen 0.529310
Wendy E. Mackay 0.200000
Philippe Caillou 0.674419
Alain Denise 0.297222
Jean-Daniel Fekete 0.437107
Emmanuel Pietriga 0.260317
Yann Ponty 0.586364
Marc Schoenauer 0.443165
Franck Cappello 0.480488
Caroline Appert 0.273913
Michel Beaudouin-Lafon 0.297531
Wendy Mackay 0.302128
Anne Auger 0.591139
Evelyne Lutton 0.272973
Pierre Dragicevic 0.317284
Ioana Manolescu 0.507317
Nikolaus Hansen 0.686420
Nicolas Bredeche 0.544118
Olivier Teytaud 0.567925
François Goasdoué 0.430189
Nathalie Pernelle 0.482353
Fatiha Saïs 0.507317
Sarah Cohen-Boulakia 0.581818
Claude Marché 0.465957
Chantal Reynaud 0.513333
Olivier Chapuis 0.384615
Steven Martin 0.356410
Fatiha Zaidi 0.390625
Balázs Kégl 0.286842
Paola Tubaro 0.602564
Raymond Ros 0.655882
Cyril Furtlehner 0.666667
Anastasia Bezerianos 0.319403
Sylvie Boldo 0.488571
Guillaume Melquiond 0.472727
Marc Baboulin 0.540000
Dimo Brockhoff 0.610256
Nathann Cohen 0.380488
Petra Isenberg 0.479439
Tobias Isenberg 0.536752
Loïc Paulevé 0.726190
Céline Gicquel 0.818421
Isabelle Guyon 0.623596
Guillaume Charpiat 0.522581
Lonni Besançon 0.487879
Name: 0, dtype: float64
================================
Run params : {}
--------------------------------
Scoring params : {}
================================
----------- Scores -----------
custom_score_lsa : 10.0000
--------- Desc Scores ---------
--------- Raw Scores ---------
custom scores:
None
================================
Run params : {'2_nD__rkey': 'lsa', '2_nD__rnum_dims': 50, '2_nD__rnormalize': True, '2_nD__rextra_param': True}
--------------------------------
Scoring params : {'2_nD__smin_score': 30, '2_nD__srecompute': True, '2_nD__ssample_size': None, '2_nD__sn_neighbors': 10, '2_nD__srandom_state': 42, '2_nD__sfactor': 10}
================================
----------- Scores -----------
2_nD__neighbors_articles_authors : 0.4779
2_nD__custom_score_lsa : 10.0000
--------- Desc Scores ---------
2_nD__neighbors_articles_authors_det
0 Céline Gicquel : 0.82
1 Loïc Paulevé : 0.73
2 Nikolaus Hansen : 0.69
3 Sébastien Tixeuil : 0.68
4 Philippe Caillou : 0.67
5 Cyril Furtlehner : 0.67
6 Raymond Ros : 0.66
7 Isabelle Guyon : 0.62
8 Dimo Brockhoff : 0.61
9 Paola Tubaro : 0.60
10 Anne Auger : 0.59
11 Yann Ponty : 0.59
12 Sarah Cohen-Boulakia : 0.58
13 Olivier Teytaud : 0.57
14 Nicolas Bredeche : 0.54
15 Marc Baboulin : 0.54
16 Tobias Isenberg : 0.54
17 Albert Cohen : 0.53
18 Guillaume Charpiat : 0.52
19 Chantal Reynaud : 0.51
20 Fatiha Saïs : 0.51
21 Ioana Manolescu : 0.51
22 Sylvie Boldo : 0.49
23 Lonni Besançon : 0.49
24 Nathalie Pernelle : 0.48
25 Franck Cappello : 0.48
26 Petra Isenberg : 0.48
27 Guillaume Melquiond : 0.47
28 Claude Marché : 0.47
29 Marc Schoenauer : 0.44
30 Jean-Daniel Fekete : 0.44
31 François Goasdoué : 0.43
32 Fatiha Zaidi : 0.39
33 Olivier Chapuis : 0.38
34 Michèle Sebag : 0.38
35 Nathann Cohen : 0.38
36 Steven Martin : 0.36
37 Anastasia Bezerianos : 0.32
38 Pierre Dragicevic : 0.32
39 Wendy Mackay : 0.30
40 Michel Beaudouin-Lafon : 0.30
41 Alain Denise : 0.30
42 Balázs Kégl : 0.29
43 Caroline Appert : 0.27
44 Evelyne Lutton : 0.27
45 Johanne Cohen : 0.27
46 Emmanuel Pietriga : 0.26
47 Wendy E. Mackay : 0.20
Name: 0, dtype: object
VALUE : 0.4779
--------- Raw Scores ---------
2_nD__neighbors_articles_authors
Sébastien Tixeuil 0.680851
Michèle Sebag 0.383942
Johanne Cohen 0.272093
Albert Cohen 0.529310
Wendy E. Mackay 0.200000
Philippe Caillou 0.674419
Alain Denise 0.297222
Jean-Daniel Fekete 0.437107
Emmanuel Pietriga 0.260317
Yann Ponty 0.586364
Marc Schoenauer 0.443165
Franck Cappello 0.480488
Caroline Appert 0.273913
Michel Beaudouin-Lafon 0.297531
Wendy Mackay 0.302128
Anne Auger 0.591139
Evelyne Lutton 0.272973
Pierre Dragicevic 0.317284
Ioana Manolescu 0.507317
Nikolaus Hansen 0.686420
Nicolas Bredeche 0.544118
Olivier Teytaud 0.567925
François Goasdoué 0.430189
Nathalie Pernelle 0.482353
Fatiha Saïs 0.507317
Sarah Cohen-Boulakia 0.581818
Claude Marché 0.465957
Chantal Reynaud 0.513333
Olivier Chapuis 0.384615
Steven Martin 0.356410
Fatiha Zaidi 0.390625
Balázs Kégl 0.286842
Paola Tubaro 0.602564
Raymond Ros 0.655882
Cyril Furtlehner 0.666667
Anastasia Bezerianos 0.319403
Sylvie Boldo 0.488571
Guillaume Melquiond 0.472727
Marc Baboulin 0.540000
Dimo Brockhoff 0.610256
Nathann Cohen 0.380488
Petra Isenberg 0.479439
Tobias Isenberg 0.536752
Loïc Paulevé 0.726190
Céline Gicquel 0.818421
Isabelle Guyon 0.623596
Guillaume Charpiat 0.522581
Lonni Besançon 0.487879
Name: 0, dtype: float64
2_projection_nD scores:
None
umap matrices generated.
================================
Run params : {'3_2D__rkey': 'umap', '3_2D__rmetric': 'euclidean', '3_2D__rn_neighbors': 50, '3_2D__rmin_dist': 0.25, '3_2D__rinit': 'random', '3_2D__rlearning_rate': 1.0, '3_2D__rn_epochs': None, '3_2D__rrandom_state': None}
--------------------------------
Scoring params : {'3_2D__smin_score': 30, '3_2D__srecompute': True, '3_2D__ssample_size': None, '3_2D__sn_neighbors': 10, '3_2D__srandom_state': 42}
================================
----------- Scores -----------
3_2D__neighbors_articles_authors : 0.3995
3_2D__2_nD_3_2D_neighbors_articles_authors_dim_mean : -0.0784
3_2D__2_nD_3_2D_neighbors_articles_authors_dim_median : -0.0771
--------- Desc Scores ---------
3_2D__neighbors_articles_authors_det
0 Loïc Paulevé : 0.68
1 Céline Gicquel : 0.67
2 Cyril Furtlehner : 0.62
3 Philippe Caillou : 0.60
4 Sébastien Tixeuil : 0.58
5 Nikolaus Hansen : 0.58
6 Isabelle Guyon : 0.53
7 Nicolas Bredeche : 0.52
8 Olivier Teytaud : 0.52
9 Anne Auger : 0.51
10 Yann Ponty : 0.51
11 Raymond Ros : 0.49
12 Dimo Brockhoff : 0.49
13 Sylvie Boldo : 0.47
14 Tobias Isenberg : 0.46
15 Marc Baboulin : 0.46
16 Albert Cohen : 0.44
17 Sarah Cohen-Boulakia : 0.43
18 Chantal Reynaud : 0.43
19 Franck Cappello : 0.42
20 Guillaume Melquiond : 0.40
21 Ioana Manolescu : 0.39
22 Paola Tubaro : 0.39
23 Nathann Cohen : 0.39
24 Jean-Daniel Fekete : 0.39
25 Marc Schoenauer : 0.39
26 Lonni Besançon : 0.38
27 Nathalie Pernelle : 0.38
28 Fatiha Saïs : 0.38
29 Steven Martin : 0.35
30 Petra Isenberg : 0.34
31 Claude Marché : 0.34
32 Michèle Sebag : 0.33
33 Guillaume Charpiat : 0.33
34 Pierre Dragicevic : 0.31
35 Olivier Chapuis : 0.31
36 Fatiha Zaidi : 0.31
37 François Goasdoué : 0.30
38 Alain Denise : 0.28
39 Michel Beaudouin-Lafon : 0.27
40 Anastasia Bezerianos : 0.27
41 Johanne Cohen : 0.25
42 Balázs Kégl : 0.22
43 Wendy Mackay : 0.22
44 Caroline Appert : 0.21
45 Wendy E. Mackay : 0.21
46 Emmanuel Pietriga : 0.21
47 Evelyne Lutton : 0.19
dtype: object
VALUE : 0.3995
3_2D__2_nD_3_2D_neighbors_articles_authors_dim_det
0 Paola Tubaro : -0.21
1 Guillaume Charpiat : -0.19
2 Raymond Ros : -0.16
3 Sarah Cohen-Boulakia : -0.15
4 Céline Gicquel : -0.15
5 Petra Isenberg : -0.14
6 François Goasdoué : -0.13
7 Fatiha Saïs : -0.13
8 Claude Marché : -0.12
9 Dimo Brockhoff : -0.12
10 Ioana Manolescu : -0.11
11 Nikolaus Hansen : -0.11
12 Lonni Besançon : -0.10
13 Nathalie Pernelle : -0.10
14 Sébastien Tixeuil : -0.10
15 Isabelle Guyon : -0.09
16 Chantal Reynaud : -0.09
17 Albert Cohen : -0.09
18 Fatiha Zaidi : -0.08
19 Wendy Mackay : -0.08
20 Anne Auger : -0.08
21 Marc Baboulin : -0.08
22 Evelyne Lutton : -0.08
23 Yann Ponty : -0.08
24 Olivier Chapuis : -0.08
25 Tobias Isenberg : -0.07
26 Guillaume Melquiond : -0.07
27 Philippe Caillou : -0.07
28 Balázs Kégl : -0.07
29 Caroline Appert : -0.06
30 Marc Schoenauer : -0.06
31 Franck Cappello : -0.06
32 Emmanuel Pietriga : -0.05
33 Anastasia Bezerianos : -0.05
34 Michèle Sebag : -0.05
35 Jean-Daniel Fekete : -0.05
36 Olivier Teytaud : -0.05
37 Cyril Furtlehner : -0.05
38 Loïc Paulevé : -0.04
39 Michel Beaudouin-Lafon : -0.03
40 Johanne Cohen : -0.03
41 Nicolas Bredeche : -0.02
42 Alain Denise : -0.02
43 Sylvie Boldo : -0.01
44 Steven Martin : -0.01
45 Pierre Dragicevic : -0.00
46 Wendy E. Mackay : 0.01
47 Nathann Cohen : 0.01
dtype: object
VALUE : -0.0784
--------- Raw Scores ---------
3_2D__neighbors_articles_authors
Sébastien Tixeuil 0.582979
Michèle Sebag 0.334307
Johanne Cohen 0.246512
Albert Cohen 0.443103
Wendy E. Mackay 0.206522
Philippe Caillou 0.604651
Alain Denise 0.277778
Jean-Daniel Fekete 0.388050
Emmanuel Pietriga 0.206349
Yann Ponty 0.509091
Marc Schoenauer 0.387050
Franck Cappello 0.424390
Caroline Appert 0.213043
Michel Beaudouin-Lafon 0.271605
Wendy Mackay 0.219149
Anne Auger 0.510127
Evelyne Lutton 0.194595
Pierre Dragicevic 0.314815
Ioana Manolescu 0.393902
Nikolaus Hansen 0.575309
Nicolas Bredeche 0.520588
Olivier Teytaud 0.519811
François Goasdoué 0.298113
Nathalie Pernelle 0.379412
Fatiha Saïs 0.378049
Sarah Cohen-Boulakia 0.427273
Claude Marché 0.342553
Chantal Reynaud 0.426667
Olivier Chapuis 0.307692
Steven Martin 0.346154
Fatiha Zaidi 0.306250
Balázs Kégl 0.221053
Paola Tubaro 0.392308
Raymond Ros 0.494118
Cyril Furtlehner 0.620513
Anastasia Bezerianos 0.265672
Sylvie Boldo 0.474286
Guillaume Melquiond 0.400000
Marc Baboulin 0.460000
Dimo Brockhoff 0.492308
Nathann Cohen 0.390244
Petra Isenberg 0.342991
Tobias Isenberg 0.463248
Loïc Paulevé 0.683333
Céline Gicquel 0.668421
Isabelle Guyon 0.533708
Guillaume Charpiat 0.332258
Lonni Besançon 0.384848
dtype: float64
3_2D__2_nD_3_2D_neighbors_articles_authors_dim
2_nD__neighbors_articles_authors \
Sébastien Tixeuil 0.680851
Michèle Sebag 0.383942
Johanne Cohen 0.272093
Albert Cohen 0.529310
Wendy E. Mackay 0.200000
Philippe Caillou 0.674419
Alain Denise 0.297222
Jean-Daniel Fekete 0.437107
Emmanuel Pietriga 0.260317
Yann Ponty 0.586364
Marc Schoenauer 0.443165
Franck Cappello 0.480488
Caroline Appert 0.273913
Michel Beaudouin-Lafon 0.297531
Wendy Mackay 0.302128
Anne Auger 0.591139
Evelyne Lutton 0.272973
Pierre Dragicevic 0.317284
Ioana Manolescu 0.507317
Nikolaus Hansen 0.686420
Nicolas Bredeche 0.544118
Olivier Teytaud 0.567925
François Goasdoué 0.430189
Nathalie Pernelle 0.482353
Fatiha Saïs 0.507317
Sarah Cohen-Boulakia 0.581818
Claude Marché 0.465957
Chantal Reynaud 0.513333
Olivier Chapuis 0.384615
Steven Martin 0.356410
Fatiha Zaidi 0.390625
Balázs Kégl 0.286842
Paola Tubaro 0.602564
Raymond Ros 0.655882
Cyril Furtlehner 0.666667
Anastasia Bezerianos 0.319403
Sylvie Boldo 0.488571
Guillaume Melquiond 0.472727
Marc Baboulin 0.540000
Dimo Brockhoff 0.610256
Nathann Cohen 0.380488
Petra Isenberg 0.479439
Tobias Isenberg 0.536752
Loïc Paulevé 0.726190
Céline Gicquel 0.818421
Isabelle Guyon 0.623596
Guillaume Charpiat 0.522581
Lonni Besançon 0.487879
3_2D__neighbors_articles_authors scdec
Sébastien Tixeuil 0.582979 -0.097872
Michèle Sebag 0.334307 -0.049635
Johanne Cohen 0.246512 -0.025581
Albert Cohen 0.443103 -0.086207
Wendy E. Mackay 0.206522 0.006522
Philippe Caillou 0.604651 -0.069767
Alain Denise 0.277778 -0.019444
Jean-Daniel Fekete 0.388050 -0.049057
Emmanuel Pietriga 0.206349 -0.053968
Yann Ponty 0.509091 -0.077273
Marc Schoenauer 0.387050 -0.056115
Franck Cappello 0.424390 -0.056098
Caroline Appert 0.213043 -0.060870
Michel Beaudouin-Lafon 0.271605 -0.025926
Wendy Mackay 0.219149 -0.082979
Anne Auger 0.510127 -0.081013
Evelyne Lutton 0.194595 -0.078378
Pierre Dragicevic 0.314815 -0.002469
Ioana Manolescu 0.393902 -0.113415
Nikolaus Hansen 0.575309 -0.111111
Nicolas Bredeche 0.520588 -0.023529
Olivier Teytaud 0.519811 -0.048113
François Goasdoué 0.298113 -0.132075
Nathalie Pernelle 0.379412 -0.102941
Fatiha Saïs 0.378049 -0.129268
Sarah Cohen-Boulakia 0.427273 -0.154545
Claude Marché 0.342553 -0.123404
Chantal Reynaud 0.426667 -0.086667
Olivier Chapuis 0.307692 -0.076923
Steven Martin 0.346154 -0.010256
Fatiha Zaidi 0.306250 -0.084375
Balázs Kégl 0.221053 -0.065789
Paola Tubaro 0.392308 -0.210256
Raymond Ros 0.494118 -0.161765
Cyril Furtlehner 0.620513 -0.046154
Anastasia Bezerianos 0.265672 -0.053731
Sylvie Boldo 0.474286 -0.014286
Guillaume Melquiond 0.400000 -0.072727
Marc Baboulin 0.460000 -0.080000
Dimo Brockhoff 0.492308 -0.117949
Nathann Cohen 0.390244 0.009756
Petra Isenberg 0.342991 -0.136449
Tobias Isenberg 0.463248 -0.073504
Loïc Paulevé 0.683333 -0.042857
Céline Gicquel 0.668421 -0.150000
Isabelle Guyon 0.533708 -0.089888
Guillaume Charpiat 0.332258 -0.190323
Lonni Besançon 0.384848 -0.103030
3_projection_2D scores:
None
Nothing in cache, initial Fitting with min_cluster_size=15 Found 69 clusters in 0.26351266100027715s
Max Fitting with min_cluster_size=30 Found 48 clusters in 0.1049684090030496s
Max Fitting with min_cluster_size=60 Found 24 clusters in 0.10055748499871697s
Max Fitting with min_cluster_size=120 Found 2 clusters in 0.10391530300330487s
Midpoint Fitting with min_cluster_size=90 Found 2 clusters in 0.10530399999697693s
Midpoint Fitting with min_cluster_size=75 Found 2 clusters in 0.10460692900232971s
Midpoint Fitting with min_cluster_size=67 Found 19 clusters in 0.10069006399862701s
No need Re-Fitting with min_cluster_size=67
Clusters cached: [2, 2, 2, 19, 24, 48, 69]
Nothing in cache, initial Fitting with min_cluster_size=15 Found 69 clusters in 0.10288854900136357s
Max Fitting with min_cluster_size=30 Found 48 clusters in 0.10291053100081626s
Max Fitting with min_cluster_size=60 Found 24 clusters in 0.09792928800015943s
No need Re-Fitting with min_cluster_size=60
Clusters cached: [24, 48, 69]
================================
Run params : {'4_clus__rkey': 'hdbscan', '4_clus__rbase_factor': 3}
--------------------------------
Scoring params : {'4_clus__siter_stab': 2, '4_clus__sremove_stab': [0, 0.01, 0.03, 0.1, 0.25], '4_clus__smetric': 'euclidean', '4_clus__srandom_state': None}
================================
----------- Scores -----------
4_clus__nb_clust_0 : 8.0000
4_clus__silhouette_0 : 0.0979
4_clus__avg_word_couv_0 : 0.6286
4_clus__med_word_couv_0 : 0.6259
4_clus__avg_word_couv_minus_0 : 0.5985
4_clus__big_small_ratio_0 : 18.0423
4_clus__stab_clus_0 : 0.0105
4_clus__nb_clust_1 : 24.0000
4_clus__silhouette_1 : 0.1217
4_clus__avg_word_couv_1 : 0.5620
4_clus__med_word_couv_1 : 0.5542
4_clus__avg_word_couv_minus_1 : 0.5339
4_clus__big_small_ratio_1 : 21.4531
4_clus__stab_clus_1 : 0.0042
4_clus__avg_stab_avg : 0.0073
4_clus__avg_couv_avg : 0.5953
4_clus__clu_score : 0.3013
--------- Desc Scores ---------
4_clus__clus_eval_pos_0_det
0 query, humanities social sciences : s 603 stb ...
1 logic science, semantics : s 270 stb 0.10 + 0....
2 neural networks, challenge : s 224 stb 0.00 + ...
3 biology, biological : s 185 stb 0.00 + 0.54 - ...
4 networking internet architecture, stabilizing ...
5 compiler, parallelism : s 153 stb 0.00 + 0.60 ...
6 creative, movement : s 145 stb 0.00 + 0.44 - 0.01
7 secondary structure, quantitative methods : s ...
8 vertices, minimum : s 131 stb 0.00 + 0.68 - 0.02
9 fluid, dynamics : s 126 stb 0.00 + 0.68 - 0.05
10 automata, combinatorics : s 124 stb 0.00 + 0.5...
11 visualizations, illustrative : s 120 stb 0.00 ...
12 services, cluster computing : s 117 stb 0.00 +...
13 optimization control, population : s 94 stb 0....
14 gestures, pointing : s 90 stb 0.00 + 0.52 - 0.01
15 reality, tangible : s 83 stb 0.00 + 0.76 - 0.01
16 operations, stochastic : s 77 stb 0.00 + 0.73 ...
17 monte carlo, games : s 74 stb 0.00 + 0.97 - 0.02
18 adaptive, multi armed : s 71 stb 0.00 + 0.90 -...
dtype: object
VALUE : 0.6286
4_clus__clus_eval_pos_1_det
0 verification, testing : s 270 stb 0.00 + 0.57 ...
1 statistics, reinforcement learning : s 224 stb...
2 molecular, protein : s 185 stb 0.00 + 0.45 - 0.02
3 networking internet, wireless : s 162 stb 0.00...
4 programming languages, architectures : s 153 s...
5 interfaces, motion : s 145 stb 0.00 + 0.44 - 0.04
6 secondary, sequence : s 132 stb 0.00 + 0.58 - ...
7 discrete mathematics, vertex : s 131 stb 0.00 ...
8 numerical simulations, mechanics : s 126 stb 0...
9 finite, trees : s 124 stb 0.00 + 0.52 - 0.06
10 documents, mining : s 120 stb 0.00 + 0.52 - 0.03
11 information visualization, representations : s...
12 service, fault : s 117 stb 0.00 + 0.68 - 0.04
13 queries, query answers : s 97 stb 0.00 + 0.78 ...
14 evolution strategies, noisy optimization : s 9...
15 corpora, natural language : s 91 stb 0.00 + 0....
16 spatial, navigation : s 90 stb 0.00 + 0.44 - 0.04
17 virtual, augmented reality : s 83 stb 0.00 + 0...
18 integer, chance constrained : s 77 stb 0.00 + ...
19 monte carlo search, carlo : s 74 stb 0.00 + 0....
20 social sciences, social networks : s 73 stb 0....
21 multi armed bandit, multi armed bandits : s 71...
22 image, similarity : s 66 stb 0.00 + 0.83 - 0.03
23 linear systems, matrices : s 64 stb 0.10 + 0.7...
dtype: object
VALUE : 0.5620
--------- Raw Scores ---------
4_clustering scores:
None
['2_nD__neighbors_articles_authors', '3_2D__neighbors_articles_authors', '4_clus__clu_score']
================================
Run params : {}
--------------------------------
Scoring params : {'6_pst__sname_list': ['2_nD__neighbors_articles_authors', '3_2D__neighbors_articles_authors', '4_clus__clu_score']}
================================
----------- Scores -----------
6_pst__final_score : 0.3929
--------- Desc Scores ---------
6_pst__final_score_det
0 2_nD__neighbors_articles_authors : 0.48
1 3_2D__neighbors_articles_authors : 0.40
2 4_clus__clu_score : 0.30
dtype: object
VALUE : 0.3929
--------- Raw Scores ---------
6_post scores:
None
Finished running step!
<cartodata.model_selection.utils.Results object at 0x7efc8769c490>
.. GENERATED FROM PYTHON SOURCE LINES 522-523
We can verify that the experiment is run for 2 new set of parameters:
.. GENERATED FROM PYTHON SOURCE LINES 523-528
.. code-block:: Python
df = pd.read_csv(TOP_DIR / "experiment.csv", index_col=0)
df
.. raw:: html
<div class="output_subarea output_html rendered_html output_result">
<div>
<style scoped>
.dataframe tbody tr th:only-of-type {
vertical-align: middle;
}
.dataframe tbody tr th {
vertical-align: top;
}
.dataframe thead th {
text-align: right;
}
</style>
<table border="1" class="dataframe">
<thead>
<tr style="text-align: right;">
<th></th>
<th>id</th>
<th>robustseed</th>
<th>authors__filter_min_score</th>
<th>filter_min_score</th>
<th>projection_nD</th>
<th>projection_2D</th>
<th>selected</th>
</tr>
</thead>
<tbody>
<tr>
<th>0</th>
<td>68d289ffd85ae917e710</td>
<td>0</td>
<td>4</td>
<td>6</td>
<td>{'key': 'lsa', 'num_dims': 50, 'extra_param': ...</td>
<td>{'key': 'umap', 'n_neighbors': 10, 'min_dist':...</td>
<td>True</td>
</tr>
<tr>
<th>1</th>
<td>ce1c0d334fc75a6570a5</td>
<td>0</td>
<td>4</td>
<td>6</td>
<td>{'key': 'lsa', 'num_dims': 50, 'extra_param': ...</td>
<td>{'key': 'umap', 'n_neighbors': 10, 'min_dist':...</td>
<td>True</td>
</tr>
<tr>
<th>2</th>
<td>a588bce09630b37c4111</td>
<td>0</td>
<td>4</td>
<td>6</td>
<td>{'key': 'lsa', 'num_dims': 50, 'extra_param': ...</td>
<td>{'key': 'umap', 'n_neighbors': 10, 'min_dist':...</td>
<td>True</td>
</tr>
<tr>
<th>3</th>
<td>06128c38fd13123e263c</td>
<td>0</td>
<td>4</td>
<td>6</td>
<td>{'key': 'lsa', 'num_dims': 50, 'extra_param': ...</td>
<td>{'key': 'umap', 'n_neighbors': 20, 'min_dist':...</td>
<td>True</td>
</tr>
<tr>
<th>4</th>
<td>1040be9e44c8ab04aa0e</td>
<td>0</td>
<td>4</td>
<td>6</td>
<td>{'key': 'lsa', 'num_dims': 50, 'extra_param': ...</td>
<td>{'key': 'umap', 'n_neighbors': 20, 'min_dist':...</td>
<td>True</td>
</tr>
<tr>
<th>5</th>
<td>014dbb63a341d2622f25</td>
<td>0</td>
<td>4</td>
<td>6</td>
<td>{'key': 'lsa', 'num_dims': 50, 'extra_param': ...</td>
<td>{'key': 'umap', 'n_neighbors': 20, 'min_dist':...</td>
<td>True</td>
</tr>
<tr>
<th>6</th>
<td>06a7ca0915d11a6cad00</td>
<td>0</td>
<td>4</td>
<td>6</td>
<td>{'key': 'lsa', 'num_dims': 50, 'extra_param': ...</td>
<td>{'key': 'umap', 'n_neighbors': 50, 'min_dist':...</td>
<td>True</td>
</tr>
<tr>
<th>7</th>
<td>979bdf3fc81db6766d24</td>
<td>0</td>
<td>4</td>
<td>6</td>
<td>{'key': 'lsa', 'num_dims': 50, 'extra_param': ...</td>
<td>{'key': 'umap', 'n_neighbors': 50, 'min_dist':...</td>
<td>True</td>
</tr>
<tr>
<th>8</th>
<td>74f9b611573d48c19a8e</td>
<td>0</td>
<td>4</td>
<td>6</td>
<td>{'key': 'lsa', 'num_dims': 50, 'extra_param': ...</td>
<td>{'key': 'umap', 'n_neighbors': 50, 'min_dist':...</td>
<td>False</td>
</tr>
<tr>
<th>9</th>
<td>c5ec62b56a0c0563c079</td>
<td>0</td>
<td>4</td>
<td>6</td>
<td>{'key': 'lsa', 'num_dims': 50, 'extra_param': ...</td>
<td>{'key': 'umap', 'n_neighbors': 10, 'min_dist':...</td>
<td>False</td>
</tr>
<tr>
<th>10</th>
<td>b11b71ed11b157e13f95</td>
<td>0</td>
<td>4</td>
<td>6</td>
<td>{'key': 'lsa', 'num_dims': 50, 'extra_param': ...</td>
<td>{'key': 'umap', 'n_neighbors': 10, 'min_dist':...</td>
<td>False</td>
</tr>
<tr>
<th>11</th>
<td>a7909c645cf88ed5c209</td>
<td>0</td>
<td>4</td>
<td>6</td>
<td>{'key': 'lsa', 'num_dims': 50, 'extra_param': ...</td>
<td>{'key': 'umap', 'n_neighbors': 10, 'min_dist':...</td>
<td>False</td>
</tr>
<tr>
<th>12</th>
<td>4c16fefcccebf0b96309</td>
<td>0</td>
<td>4</td>
<td>6</td>
<td>{'key': 'lsa', 'num_dims': 50, 'extra_param': ...</td>
<td>{'key': 'umap', 'n_neighbors': 20, 'min_dist':...</td>
<td>False</td>
</tr>
<tr>
<th>13</th>
<td>97a0b8354402e18fa744</td>
<td>0</td>
<td>4</td>
<td>6</td>
<td>{'key': 'lsa', 'num_dims': 50, 'extra_param': ...</td>
<td>{'key': 'umap', 'n_neighbors': 20, 'min_dist':...</td>
<td>False</td>
</tr>
<tr>
<th>14</th>
<td>dcfa478890363f0396c3</td>
<td>0</td>
<td>4</td>
<td>6</td>
<td>{'key': 'lsa', 'num_dims': 50, 'extra_param': ...</td>
<td>{'key': 'umap', 'n_neighbors': 20, 'min_dist':...</td>
<td>False</td>
</tr>
<tr>
<th>15</th>
<td>0070d0c653c88c823b5c</td>
<td>0</td>
<td>4</td>
<td>6</td>
<td>{'key': 'lsa', 'num_dims': 50, 'extra_param': ...</td>
<td>{'key': 'umap', 'n_neighbors': 50, 'min_dist':...</td>
<td>False</td>
</tr>
<tr>
<th>16</th>
<td>9fd981e9c09b2eed1de6</td>
<td>0</td>
<td>4</td>
<td>6</td>
<td>{'key': 'lsa', 'num_dims': 50, 'extra_param': ...</td>
<td>{'key': 'umap', 'n_neighbors': 50, 'min_dist':...</td>
<td>False</td>
</tr>
<tr>
<th>17</th>
<td>8ff3e42500d4389f445d</td>
<td>0</td>
<td>4</td>
<td>6</td>
<td>{'key': 'lsa', 'num_dims': 50, 'extra_param': ...</td>
<td>{'key': 'umap', 'n_neighbors': 50, 'min_dist':...</td>
<td>False</td>
</tr>
<tr>
<th>18</th>
<td>9bd0b3935ca3e4457d8a</td>
<td>0</td>
<td>4</td>
<td>6</td>
<td>{'key': 'lsa', 'num_dims': 100, 'extra_param':...</td>
<td>{'key': 'umap', 'n_neighbors': 10, 'min_dist':...</td>
<td>False</td>
</tr>
<tr>
<th>19</th>
<td>9a6fd2af303ac5430413</td>
<td>0</td>
<td>4</td>
<td>6</td>
<td>{'key': 'lsa', 'num_dims': 100, 'extra_param':...</td>
<td>{'key': 'umap', 'n_neighbors': 10, 'min_dist':...</td>
<td>False</td>
</tr>
<tr>
<th>20</th>
<td>151ba7fff2224cb7d57b</td>
<td>0</td>
<td>4</td>
<td>6</td>
<td>{'key': 'lsa', 'num_dims': 100, 'extra_param':...</td>
<td>{'key': 'umap', 'n_neighbors': 10, 'min_dist':...</td>
<td>False</td>
</tr>
<tr>
<th>21</th>
<td>45e2d7c0f5ed87a9653b</td>
<td>0</td>
<td>4</td>
<td>6</td>
<td>{'key': 'lsa', 'num_dims': 100, 'extra_param':...</td>
<td>{'key': 'umap', 'n_neighbors': 20, 'min_dist':...</td>
<td>False</td>
</tr>
<tr>
<th>22</th>
<td>fb97932fdc7a24071f48</td>
<td>0</td>
<td>4</td>
<td>6</td>
<td>{'key': 'lsa', 'num_dims': 100, 'extra_param':...</td>
<td>{'key': 'umap', 'n_neighbors': 20, 'min_dist':...</td>
<td>False</td>
</tr>
<tr>
<th>23</th>
<td>3381a66259bc4e6d8899</td>
<td>0</td>
<td>4</td>
<td>6</td>
<td>{'key': 'lsa', 'num_dims': 100, 'extra_param':...</td>
<td>{'key': 'umap', 'n_neighbors': 20, 'min_dist':...</td>
<td>False</td>
</tr>
<tr>
<th>24</th>
<td>5c605c552b449a76cbbf</td>
<td>0</td>
<td>4</td>
<td>6</td>
<td>{'key': 'lsa', 'num_dims': 100, 'extra_param':...</td>
<td>{'key': 'umap', 'n_neighbors': 50, 'min_dist':...</td>
<td>False</td>
</tr>
<tr>
<th>25</th>
<td>eae12b9b57e5abbf5b3b</td>
<td>0</td>
<td>4</td>
<td>6</td>
<td>{'key': 'lsa', 'num_dims': 100, 'extra_param':...</td>
<td>{'key': 'umap', 'n_neighbors': 50, 'min_dist':...</td>
<td>False</td>
</tr>
<tr>
<th>26</th>
<td>1595a61c3d93a94b121e</td>
<td>0</td>
<td>4</td>
<td>6</td>
<td>{'key': 'lsa', 'num_dims': 100, 'extra_param':...</td>
<td>{'key': 'umap', 'n_neighbors': 50, 'min_dist':...</td>
<td>False</td>
</tr>
<tr>
<th>27</th>
<td>d1ced9d06c4ed06937ad</td>
<td>0</td>
<td>4</td>
<td>6</td>
<td>{'key': 'lsa', 'num_dims': 100, 'extra_param':...</td>
<td>{'key': 'umap', 'n_neighbors': 10, 'min_dist':...</td>
<td>False</td>
</tr>
<tr>
<th>28</th>
<td>fc5671119a7e46e9b519</td>
<td>0</td>
<td>4</td>
<td>6</td>
<td>{'key': 'lsa', 'num_dims': 100, 'extra_param':...</td>
<td>{'key': 'umap', 'n_neighbors': 10, 'min_dist':...</td>
<td>False</td>
</tr>
<tr>
<th>29</th>
<td>c6c7cfaf3dab327815a3</td>
<td>0</td>
<td>4</td>
<td>6</td>
<td>{'key': 'lsa', 'num_dims': 100, 'extra_param':...</td>
<td>{'key': 'umap', 'n_neighbors': 10, 'min_dist':...</td>
<td>False</td>
</tr>
<tr>
<th>30</th>
<td>e3c9335d86928f04d986</td>
<td>0</td>
<td>4</td>
<td>6</td>
<td>{'key': 'lsa', 'num_dims': 100, 'extra_param':...</td>
<td>{'key': 'umap', 'n_neighbors': 20, 'min_dist':...</td>
<td>False</td>
</tr>
<tr>
<th>31</th>
<td>b2b14d84c6cf741b3f94</td>
<td>0</td>
<td>4</td>
<td>6</td>
<td>{'key': 'lsa', 'num_dims': 100, 'extra_param':...</td>
<td>{'key': 'umap', 'n_neighbors': 20, 'min_dist':...</td>
<td>False</td>
</tr>
<tr>
<th>32</th>
<td>e079875025cff951a37d</td>
<td>0</td>
<td>4</td>
<td>6</td>
<td>{'key': 'lsa', 'num_dims': 100, 'extra_param':...</td>
<td>{'key': 'umap', 'n_neighbors': 20, 'min_dist':...</td>
<td>False</td>
</tr>
<tr>
<th>33</th>
<td>943a271c14a11b4e5954</td>
<td>0</td>
<td>4</td>
<td>6</td>
<td>{'key': 'lsa', 'num_dims': 100, 'extra_param':...</td>
<td>{'key': 'umap', 'n_neighbors': 50, 'min_dist':...</td>
<td>False</td>
</tr>
<tr>
<th>34</th>
<td>fea341b317d4d596ea82</td>
<td>0</td>
<td>4</td>
<td>6</td>
<td>{'key': 'lsa', 'num_dims': 100, 'extra_param':...</td>
<td>{'key': 'umap', 'n_neighbors': 50, 'min_dist':...</td>
<td>False</td>
</tr>
<tr>
<th>35</th>
<td>ba4951c2c78e0031fb8e</td>
<td>0</td>
<td>4</td>
<td>6</td>
<td>{'key': 'lsa', 'num_dims': 100, 'extra_param':...</td>
<td>{'key': 'umap', 'n_neighbors': 50, 'min_dist':...</td>
<td>False</td>
</tr>
<tr>
<th>36</th>
<td>939f3d80d6ec35db69f7</td>
<td>0</td>
<td>4</td>
<td>6</td>
<td>{'key': 'bert', 'family': 'all-MiniLM-L6-v2'}</td>
<td>{'key': 'umap', 'n_neighbors': 10, 'min_dist':...</td>
<td>False</td>
</tr>
<tr>
<th>37</th>
<td>18a385a32e9a8250cfda</td>
<td>0</td>
<td>4</td>
<td>6</td>
<td>{'key': 'bert', 'family': 'all-MiniLM-L6-v2'}</td>
<td>{'key': 'umap', 'n_neighbors': 10, 'min_dist':...</td>
<td>False</td>
</tr>
<tr>
<th>38</th>
<td>41f146808f2ef8b013be</td>
<td>0</td>
<td>4</td>
<td>6</td>
<td>{'key': 'bert', 'family': 'all-MiniLM-L6-v2'}</td>
<td>{'key': 'umap', 'n_neighbors': 10, 'min_dist':...</td>
<td>False</td>
</tr>
<tr>
<th>39</th>
<td>f32ac02bc42412221a98</td>
<td>0</td>
<td>4</td>
<td>6</td>
<td>{'key': 'bert', 'family': 'all-MiniLM-L6-v2'}</td>
<td>{'key': 'umap', 'n_neighbors': 20, 'min_dist':...</td>
<td>False</td>
</tr>
<tr>
<th>40</th>
<td>ed7b6ebc7f1b57baa10c</td>
<td>0</td>
<td>4</td>
<td>6</td>
<td>{'key': 'bert', 'family': 'all-MiniLM-L6-v2'}</td>
<td>{'key': 'umap', 'n_neighbors': 20, 'min_dist':...</td>
<td>False</td>
</tr>
<tr>
<th>41</th>
<td>9616f8655120a922eb8f</td>
<td>0</td>
<td>4</td>
<td>6</td>
<td>{'key': 'bert', 'family': 'all-MiniLM-L6-v2'}</td>
<td>{'key': 'umap', 'n_neighbors': 20, 'min_dist':...</td>
<td>False</td>
</tr>
<tr>
<th>42</th>
<td>a72244820858b67def57</td>
<td>0</td>
<td>4</td>
<td>6</td>
<td>{'key': 'bert', 'family': 'all-MiniLM-L6-v2'}</td>
<td>{'key': 'umap', 'n_neighbors': 50, 'min_dist':...</td>
<td>False</td>
</tr>
<tr>
<th>43</th>
<td>3ae9168b3636e9c9ab4f</td>
<td>0</td>
<td>4</td>
<td>6</td>
<td>{'key': 'bert', 'family': 'all-MiniLM-L6-v2'}</td>
<td>{'key': 'umap', 'n_neighbors': 50, 'min_dist':...</td>
<td>False</td>
</tr>
<tr>
<th>44</th>
<td>ec8c846300e8221b9a52</td>
<td>0</td>
<td>4</td>
<td>6</td>
<td>{'key': 'bert', 'family': 'all-MiniLM-L6-v2'}</td>
<td>{'key': 'umap', 'n_neighbors': 50, 'min_dist':...</td>
<td>False</td>
</tr>
</tbody>
</table>
</div>
</div>
<br />
<br />
.. GENERATED FROM PYTHON SOURCE LINES 529-530
We can also view the results file:
.. GENERATED FROM PYTHON SOURCE LINES 530-539
.. code-block:: Python
df = pd.read_csv(TOP_DIR / "scores/final_results.csv", index_col=0)
df
""
df = pd.read_csv(TOP_DIR / "scores/2_nD__neighbors_articles_authors.csv")
df
.. raw:: html
<div class="output_subarea output_html rendered_html output_result">
<div>
<style scoped>
.dataframe tbody tr th:only-of-type {
vertical-align: middle;
}
.dataframe tbody tr th {
vertical-align: top;
}
.dataframe thead th {
text-align: right;
}
</style>
<table border="1" class="dataframe">
<thead>
<tr style="text-align: right;">
<th></th>
<th>Unnamed: 0</th>
<th>2_nD__neighbors_articles_authors</th>
</tr>
</thead>
<tbody>
<tr>
<th>0</th>
<td>{'id': '68d289ffd85ae917e710', '2_nD__rkey': '...</td>
<td>0.477852</td>
</tr>
<tr>
<th>1</th>
<td>{'id': 'ce1c0d334fc75a6570a5', '2_nD__rkey': '...</td>
<td>0.477852</td>
</tr>
<tr>
<th>2</th>
<td>{'id': 'a588bce09630b37c4111', '2_nD__rkey': '...</td>
<td>0.477852</td>
</tr>
<tr>
<th>3</th>
<td>{'id': '06128c38fd13123e263c', '2_nD__rkey': '...</td>
<td>0.477852</td>
</tr>
<tr>
<th>4</th>
<td>{'id': '1040be9e44c8ab04aa0e', '2_nD__rkey': '...</td>
<td>0.477852</td>
</tr>
<tr>
<th>5</th>
<td>{'id': '014dbb63a341d2622f25', '2_nD__rkey': '...</td>
<td>0.477852</td>
</tr>
<tr>
<th>6</th>
<td>{'id': '06a7ca0915d11a6cad00', '2_nD__rkey': '...</td>
<td>0.477852</td>
</tr>
<tr>
<th>7</th>
<td>{'id': '979bdf3fc81db6766d24', '2_nD__rkey': '...</td>
<td>0.477852</td>
</tr>
</tbody>
</table>
</div>
</div>
<br />
<br />
.. rst-class:: sphx-glr-timing
**Total running time of the script:** (5 minutes 16.483 seconds)
.. _sphx_glr_download_auto_examples_experiment_custom_lisn.py:
.. only:: html
.. container:: sphx-glr-footer sphx-glr-footer-example
.. container:: sphx-glr-download sphx-glr-download-jupyter
:download:`Download Jupyter notebook: experiment_custom_lisn.ipynb <experiment_custom_lisn.ipynb>`
.. container:: sphx-glr-download sphx-glr-download-python
:download:`Download Python source code: experiment_custom_lisn.py <experiment_custom_lisn.py>`
.. container:: sphx-glr-download sphx-glr-download-zip
:download:`Download zipped: experiment_custom_lisn.zip <experiment_custom_lisn.zip>`
.. only:: html
.. rst-class:: sphx-glr-signature
`Gallery generated by Sphinx-Gallery <https://sphinx-gallery.github.io>`_