selective 1.1.2__tar.gz → 1.2.0__tar.gz
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- selective-1.2.0/LICENSE.md +173 -0
- selective-1.2.0/MANIFEST.in +1 -0
- selective-1.2.0/NOTICES +16 -0
- selective-1.2.0/PKG-INFO +250 -0
- {selective-1.1.2 → selective-1.2.0}/README.md +23 -6
- {selective-1.1.2 → selective-1.2.0}/feature/_version.py +2 -2
- {selective-1.1.2 → selective-1.2.0}/feature/base.py +1 -1
- {selective-1.1.2 → selective-1.2.0}/feature/correlation.py +1 -1
- selective-1.2.0/feature/kl_divergence.py +66 -0
- {selective-1.1.2 → selective-1.2.0}/feature/linear.py +1 -1
- {selective-1.1.2 → selective-1.2.0}/feature/selector.py +10 -3
- {selective-1.1.2 → selective-1.2.0}/feature/statistical.py +17 -8
- {selective-1.1.2 → selective-1.2.0}/feature/text_based.py +1 -1
- {selective-1.1.2 → selective-1.2.0}/feature/tree_based.py +1 -1
- {selective-1.1.2 → selective-1.2.0}/feature/utils.py +1 -1
- {selective-1.1.2 → selective-1.2.0}/feature/variance.py +1 -1
- selective-1.2.0/requirements.txt +11 -0
- selective-1.2.0/selective.egg-info/PKG-INFO +250 -0
- selective-1.2.0/selective.egg-info/SOURCES.txt +38 -0
- {selective-1.1.2 → selective-1.2.0}/setup.py +4 -4
- selective-1.2.0/tests/test_base.py +23 -0
- selective-1.2.0/tests/test_benchmark.py +214 -0
- selective-1.2.0/tests/test_correlation.py +77 -0
- selective-1.2.0/tests/test_linear.py +160 -0
- selective-1.2.0/tests/test_parallel.py +162 -0
- selective-1.2.0/tests/test_stat_anova.py +113 -0
- selective-1.2.0/tests/test_stat_chi.py +79 -0
- selective-1.2.0/tests/test_stat_kl.py +95 -0
- selective-1.2.0/tests/test_stat_maximal.py +116 -0
- selective-1.2.0/tests/test_stat_mutual.py +113 -0
- selective-1.2.0/tests/test_stat_vif.py +87 -0
- selective-1.2.0/tests/test_text.py +699 -0
- selective-1.2.0/tests/test_tree.py +329 -0
- selective-1.2.0/tests/test_utils.py +29 -0
- selective-1.2.0/tests/test_variance.py +86 -0
- selective-1.1.2/PKG-INFO +0 -212
- selective-1.1.2/selective.egg-info/PKG-INFO +0 -212
- selective-1.1.2/selective.egg-info/SOURCES.txt +0 -18
- {selective-1.1.2 → selective-1.2.0}/feature/__init__.py +0 -0
- {selective-1.1.2 → selective-1.2.0}/selective.egg-info/dependency_links.txt +0 -0
- {selective-1.1.2 → selective-1.2.0}/selective.egg-info/requires.txt +0 -0
- {selective-1.1.2 → selective-1.2.0}/selective.egg-info/top_level.txt +0 -0
- {selective-1.1.2 → selective-1.2.0}/setup.cfg +0 -0
|
@@ -0,0 +1,173 @@
|
|
|
1
|
+
# Apache License
|
|
2
|
+
|
|
3
|
+
Version 2.0, January 2004
|
|
4
|
+
|
|
5
|
+
<http://www.apache.org/licenses/>
|
|
6
|
+
|
|
7
|
+
## TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION
|
|
8
|
+
|
|
9
|
+
### 1. Definitions.
|
|
10
|
+
|
|
11
|
+
"License" shall mean the terms and conditions for use, reproduction, and
|
|
12
|
+
distribution as defined by Sections 1 through 9 of this document.
|
|
13
|
+
|
|
14
|
+
"Licensor" shall mean the copyright owner or entity authorized by the copyright
|
|
15
|
+
owner that is granting the License.
|
|
16
|
+
|
|
17
|
+
"Legal Entity" shall mean the union of the acting entity and all other entities
|
|
18
|
+
that control, are controlled by, or are under common control with that entity.
|
|
19
|
+
For the purposes of this definition, "control" means (i) the power, direct or
|
|
20
|
+
indirect, to cause the direction or management of such entity, whether by
|
|
21
|
+
contract or otherwise, or (ii) ownership of fifty percent (50%) or more of the
|
|
22
|
+
outstanding shares, or (iii) beneficial ownership of such entity.
|
|
23
|
+
|
|
24
|
+
"You" (or "Your") shall mean an individual or Legal Entity exercising
|
|
25
|
+
permissions granted by this License.
|
|
26
|
+
|
|
27
|
+
"Source" form shall mean the preferred form for making modifications, including
|
|
28
|
+
but not limited to software source code, documentation source, and
|
|
29
|
+
configuration files.
|
|
30
|
+
|
|
31
|
+
"Object" form shall mean any form resulting from mechanical transformation or
|
|
32
|
+
translation of a Source form, including but not limited to compiled object
|
|
33
|
+
code, generated documentation, and conversions to other media types.
|
|
34
|
+
|
|
35
|
+
"Work" shall mean the work of authorship, whether in Source or Object form,
|
|
36
|
+
made available under the License, as indicated by a copyright notice that is
|
|
37
|
+
included in or attached to the work (an example is provided in the Appendix
|
|
38
|
+
below).
|
|
39
|
+
|
|
40
|
+
"Derivative Works" shall mean any work, whether in Source or Object form, that
|
|
41
|
+
is based on (or derived from) the Work and for which the editorial revisions,
|
|
42
|
+
annotations, elaborations, or other modifications represent, as a whole, an
|
|
43
|
+
original work of authorship. For the purposes of this License, Derivative Works
|
|
44
|
+
shall not include works that remain separable from, or merely link (or bind by
|
|
45
|
+
name) to the interfaces of, the Work and Derivative Works thereof.
|
|
46
|
+
|
|
47
|
+
"Contribution" shall mean any work of authorship, including the original
|
|
48
|
+
version of the Work and any modifications or additions to that Work or
|
|
49
|
+
Derivative Works thereof, that is intentionally submitted to Licensor for
|
|
50
|
+
inclusion in the Work by the copyright owner or by an individual or Legal
|
|
51
|
+
Entity authorized to submit on behalf of the copyright owner. For the purposes
|
|
52
|
+
of this definition, "submitted" means any form of electronic, verbal, or
|
|
53
|
+
written communication sent to the Licensor or its representatives, including
|
|
54
|
+
but not limited to communication on electronic mailing lists, source code
|
|
55
|
+
control systems, and issue tracking systems that are managed by, or on behalf
|
|
56
|
+
of, the Licensor for the purpose of discussing and improving the Work, but
|
|
57
|
+
excluding communication that is conspicuously marked or otherwise designated in
|
|
58
|
+
writing by the copyright owner as "Not a Contribution."
|
|
59
|
+
|
|
60
|
+
"Contributor" shall mean Licensor and any individual or Legal Entity on behalf
|
|
61
|
+
of whom a Contribution has been received by Licensor and subsequently
|
|
62
|
+
incorporated within the Work.
|
|
63
|
+
|
|
64
|
+
### 2. Grant of Copyright License.
|
|
65
|
+
|
|
66
|
+
Subject to the terms and conditions of this License, each Contributor hereby
|
|
67
|
+
grants to You a perpetual, worldwide, non-exclusive, no-charge, royalty-free,
|
|
68
|
+
irrevocable copyright license to reproduce, prepare Derivative Works of,
|
|
69
|
+
publicly display, publicly perform, sublicense, and distribute the Work and
|
|
70
|
+
such Derivative Works in Source or Object form.
|
|
71
|
+
|
|
72
|
+
### 3. Grant of Patent License.
|
|
73
|
+
|
|
74
|
+
Subject to the terms and conditions of this License, each Contributor hereby
|
|
75
|
+
grants to You a perpetual, worldwide, non-exclusive, no-charge, royalty-free,
|
|
76
|
+
irrevocable (except as stated in this section) patent license to make, have
|
|
77
|
+
made, use, offer to sell, sell, import, and otherwise transfer the Work, where
|
|
78
|
+
such license applies only to those patent claims licensable by such Contributor
|
|
79
|
+
that are necessarily infringed by their Contribution(s) alone or by combination
|
|
80
|
+
of their Contribution(s) with the Work to which such Contribution(s) was
|
|
81
|
+
submitted. If You institute patent litigation against any entity (including a
|
|
82
|
+
cross-claim or counterclaim in a lawsuit) alleging that the Work or a
|
|
83
|
+
Contribution incorporated within the Work constitutes direct or contributory
|
|
84
|
+
patent infringement, then any patent licenses granted to You under this License
|
|
85
|
+
for that Work shall terminate as of the date such litigation is filed.
|
|
86
|
+
|
|
87
|
+
### 4. Redistribution.
|
|
88
|
+
|
|
89
|
+
You may reproduce and distribute copies of the Work or Derivative Works thereof
|
|
90
|
+
in any medium, with or without modifications, and in Source or Object form,
|
|
91
|
+
provided that You meet the following conditions:
|
|
92
|
+
|
|
93
|
+
a. You must give any other recipients of the Work or Derivative Works a copy of
|
|
94
|
+
this License; and
|
|
95
|
+
|
|
96
|
+
b. You must cause any modified files to carry prominent
|
|
97
|
+
notices stating that You changed the files; and
|
|
98
|
+
|
|
99
|
+
c. You must retain, in the
|
|
100
|
+
Source form of any Derivative Works that You distribute, all copyright, patent,
|
|
101
|
+
trademark, and attribution notices from the Source form of the Work, excluding
|
|
102
|
+
those notices that do not pertain to any part of the Derivative Works; and
|
|
103
|
+
|
|
104
|
+
d. If the Work includes a "NOTICE" text file as part of its distribution, then
|
|
105
|
+
any Derivative Works that You distribute must include a readable copy of the
|
|
106
|
+
attribution notices contained within such NOTICE file, excluding those notices
|
|
107
|
+
that do not pertain to any part of the Derivative Works, in at least one of the
|
|
108
|
+
following places: within a NOTICE text file distributed as part of the
|
|
109
|
+
Derivative Works; within the Source form or documentation, if provided along
|
|
110
|
+
with the Derivative Works; or, within a display generated by the Derivative
|
|
111
|
+
Works, if and wherever such third-party notices normally appear. The contents
|
|
112
|
+
of the NOTICE file are for informational purposes only and do not modify the
|
|
113
|
+
License. You may add Your own attribution notices within Derivative Works that
|
|
114
|
+
You distribute, alongside or as an addendum to the NOTICE text from the Work,
|
|
115
|
+
provided that such additional attribution notices cannot be construed as
|
|
116
|
+
modifying the License. You may add Your own copyright statement to Your
|
|
117
|
+
modifications and may provide additional or different license terms and
|
|
118
|
+
conditions for use, reproduction, or distribution of Your modifications, or for
|
|
119
|
+
any such Derivative Works as a whole, provided Your use, reproduction, and
|
|
120
|
+
distribution of the Work otherwise complies with the conditions stated in this
|
|
121
|
+
License.
|
|
122
|
+
|
|
123
|
+
### 5. Submission of Contributions.
|
|
124
|
+
|
|
125
|
+
Unless You explicitly state otherwise, any Contribution intentionally submitted
|
|
126
|
+
for inclusion in the Work by You to the Licensor shall be under the terms and
|
|
127
|
+
conditions of this License, without any additional terms or conditions.
|
|
128
|
+
Notwithstanding the above, nothing herein shall supersede or modify the terms
|
|
129
|
+
of any separate license agreement you may have executed with Licensor regarding
|
|
130
|
+
such Contributions.
|
|
131
|
+
|
|
132
|
+
### 6. Trademarks.
|
|
133
|
+
|
|
134
|
+
This License does not grant permission to use the trade names, trademarks,
|
|
135
|
+
service marks, or product names of the Licensor, except as required for
|
|
136
|
+
reasonable and customary use in describing the origin of the Work and
|
|
137
|
+
reproducing the content of the NOTICE file.
|
|
138
|
+
|
|
139
|
+
### 7. Disclaimer of Warranty.
|
|
140
|
+
|
|
141
|
+
Unless required by applicable law or agreed to in writing, Licensor provides
|
|
142
|
+
the Work (and each Contributor provides its Contributions) on an "AS IS" BASIS,
|
|
143
|
+
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied,
|
|
144
|
+
including, without limitation, any warranties or conditions of TITLE,
|
|
145
|
+
NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A PARTICULAR PURPOSE. You are
|
|
146
|
+
solely responsible for determining the appropriateness of using or
|
|
147
|
+
redistributing the Work and assume any risks associated with Your exercise of
|
|
148
|
+
permissions under this License.
|
|
149
|
+
|
|
150
|
+
### 8. Limitation of Liability.
|
|
151
|
+
|
|
152
|
+
In no event and under no legal theory, whether in tort (including negligence),
|
|
153
|
+
contract, or otherwise, unless required by applicable law (such as deliberate
|
|
154
|
+
and grossly negligent acts) or agreed to in writing, shall any Contributor be
|
|
155
|
+
liable to You for damages, including any direct, indirect, special, incidental,
|
|
156
|
+
or consequential damages of any character arising as a result of this License
|
|
157
|
+
or out of the use or inability to use the Work (including but not limited to
|
|
158
|
+
damages for loss of goodwill, work stoppage, computer failure or malfunction,
|
|
159
|
+
or any and all other commercial damages or losses), even if such Contributor
|
|
160
|
+
has been advised of the possibility of such damages.
|
|
161
|
+
|
|
162
|
+
### 9. Accepting Warranty or Additional Liability.
|
|
163
|
+
|
|
164
|
+
While redistributing the Work or Derivative Works thereof, You may choose to
|
|
165
|
+
offer, and charge a fee for, acceptance of support, warranty, indemnity, or
|
|
166
|
+
other liability obligations and/or rights consistent with this License.
|
|
167
|
+
However, in accepting such obligations, You may act only on Your own behalf and
|
|
168
|
+
on Your sole responsibility, not on behalf of any other Contributor, and only
|
|
169
|
+
if You agree to indemnify, defend, and hold each Contributor harmless for any
|
|
170
|
+
liability incurred by, or claims asserted against, such Contributor by reason
|
|
171
|
+
of your accepting any such warranty or additional liability.
|
|
172
|
+
|
|
173
|
+
END OF TERMS AND CONDITIONS
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
include requirements.txt
|
selective-1.2.0/NOTICES
ADDED
|
@@ -0,0 +1,16 @@
|
|
|
1
|
+
Selective
|
|
2
|
+
Copyright 2020 and onwards FMR LLC.
|
|
3
|
+
|
|
4
|
+
This product includes software developed at
|
|
5
|
+
FMR LLC (https://www.fidelity.com/).
|
|
6
|
+
|
|
7
|
+
This product relies on the following works (and the dependencies thereof), installed separately:
|
|
8
|
+
- catboost | apache 2 | https://github.com/catboost/catboost
|
|
9
|
+
- lightgbm | mit | https://github.com/microsoft/LightGBM
|
|
10
|
+
- minepy | gpl-3.0-only | https://github.com/minepy/minepy
|
|
11
|
+
- numpy | bsd-3 | https://github.com/numpy/numpy
|
|
12
|
+
- pandas | bsd-3 | https://github.com/pandas-dev/pandas
|
|
13
|
+
- scikit-learn | bsd-3 |https://github.com/scikit-learn/scikit-learn
|
|
14
|
+
- seaborn | bsd-3 | https://github.com/mwaskom/seaborn
|
|
15
|
+
- statsmodels | bsd-3 | https://github.com/statsmodels/statsmodels
|
|
16
|
+
- xgboost | apache-2.0 | https://github.com/dmlc/xgboost
|
selective-1.2.0/PKG-INFO
ADDED
|
@@ -0,0 +1,250 @@
|
|
|
1
|
+
Metadata-Version: 2.4
|
|
2
|
+
Name: selective
|
|
3
|
+
Version: 1.2.0
|
|
4
|
+
Summary: feature selection library
|
|
5
|
+
Home-page: https://github.com/fidelity/selective
|
|
6
|
+
Author: FMR LLC
|
|
7
|
+
Project-URL: Source, https://github.com/fidelity/selective
|
|
8
|
+
Classifier: License :: OSI Approved :: Apache Software License
|
|
9
|
+
Classifier: Programming Language :: Python :: 3.8
|
|
10
|
+
Classifier: Operating System :: OS Independent
|
|
11
|
+
Requires-Python: >=3.8
|
|
12
|
+
Description-Content-Type: text/markdown
|
|
13
|
+
License-File: LICENSE.md
|
|
14
|
+
License-File: NOTICES
|
|
15
|
+
Requires-Dist: catboost
|
|
16
|
+
Requires-Dist: joblib
|
|
17
|
+
Requires-Dist: lightgbm
|
|
18
|
+
Requires-Dist: numpy
|
|
19
|
+
Requires-Dist: mip
|
|
20
|
+
Requires-Dist: pandas
|
|
21
|
+
Requires-Dist: scikit-learn
|
|
22
|
+
Requires-Dist: seaborn
|
|
23
|
+
Requires-Dist: statsmodels
|
|
24
|
+
Requires-Dist: textwiser
|
|
25
|
+
Requires-Dist: xgboost
|
|
26
|
+
Dynamic: author
|
|
27
|
+
Dynamic: classifier
|
|
28
|
+
Dynamic: description
|
|
29
|
+
Dynamic: description-content-type
|
|
30
|
+
Dynamic: home-page
|
|
31
|
+
Dynamic: license-file
|
|
32
|
+
Dynamic: project-url
|
|
33
|
+
Dynamic: requires-dist
|
|
34
|
+
Dynamic: requires-python
|
|
35
|
+
Dynamic: summary
|
|
36
|
+
|
|
37
|
+
[](https://github.com/fidelity/selective/actions/workflows/ci.yml) [](https://pypi.python.org/pypi/selective/) [](https://pypi.python.org/pypi/selective/) [](http://makeapullrequest.com) [](https://pepy.tech/project/selective)
|
|
38
|
+
|
|
39
|
+
|
|
40
|
+
# Selective: Feature Selection Library
|
|
41
|
+
**Selective** is a white-box feature selection library that supports supervised and unsupervised selection methods for classification and regression tasks.
|
|
42
|
+
|
|
43
|
+
The library provides:
|
|
44
|
+
|
|
45
|
+
* Simple to complex selection methods: Variance, Correlation, Statistical, Linear, Tree-based, or Customized.
|
|
46
|
+
* [Text-based selection](#text-based-selection) to maximize diversity in text embeddings and metadata coverage.
|
|
47
|
+
* Interoperable with data frames as the input.
|
|
48
|
+
* Automated task detection. No need to know what feature selection method works with what machine learning task.
|
|
49
|
+
* Benchmarking multiple selectors using cross-validation with built-in parallelization.
|
|
50
|
+
* Inspection of the results and feature importance.
|
|
51
|
+
|
|
52
|
+
Selective also provides optimized item selection based on diversity of text embeddings via [TextWiser](https://github.com/fidelity/textwiser) and
|
|
53
|
+
coverage of binary labels via multi-objective optimization ([AMAI'24](https://trebuchet.public.springernature.app/get_content/2c9eb6df-5c2b-42bc-89d6-4e3eb8bc8799?utm_source=rct_congratemailt&utm_medium=email&utm_campaign=nonoa_20240405&utm_content=10.1007/s10472-024-09941-x), [CPAIOR'21](https://link.springer.com/chapter/10.1007/978-3-030-78230-6_27), [DSO@IJCAI'22](https://arxiv.org/abs/2112.03105)). This approach speeds-up online experimentation and boosts recommender systems significantly as presented at [NVIDIA GTC'22](https://www.youtube.com/watch?v=_v-B2nRy79w).
|
|
54
|
+
|
|
55
|
+
Selective is developed by the Artificial Intelligence Center of Excellence at Fidelity Investments.
|
|
56
|
+
|
|
57
|
+
## Quick Start
|
|
58
|
+
```python
|
|
59
|
+
# Import Selective and SelectionMethod
|
|
60
|
+
from sklearn.datasets import fetch_california_housing
|
|
61
|
+
from feature.utils import get_data_label
|
|
62
|
+
from feature.selector import Selective, SelectionMethod
|
|
63
|
+
|
|
64
|
+
# Data
|
|
65
|
+
data, label = get_data_label(fetch_california_housing())
|
|
66
|
+
|
|
67
|
+
# Feature selectors from simple to more complex
|
|
68
|
+
selector = Selective(SelectionMethod.Variance(threshold=0.0))
|
|
69
|
+
selector = Selective(SelectionMethod.Correlation(threshold=0.5, method="pearson"))
|
|
70
|
+
selector = Selective(SelectionMethod.Statistical(num_features=3, method="anova"))
|
|
71
|
+
selector = Selective(SelectionMethod.Linear(num_features=3, regularization="none"))
|
|
72
|
+
selector = Selective(SelectionMethod.TreeBased(num_features=3))
|
|
73
|
+
|
|
74
|
+
# Feature reduction
|
|
75
|
+
subset = selector.fit_transform(data, label)
|
|
76
|
+
print("Reduction:", list(subset.columns))
|
|
77
|
+
print("Scores:", list(selector.get_absolute_scores()))
|
|
78
|
+
```
|
|
79
|
+
|
|
80
|
+
|
|
81
|
+
## Available Methods
|
|
82
|
+
|
|
83
|
+
| Method | Options |
|
|
84
|
+
|:--------------------------------------------------------------------------------------------------------------------------:|:--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------:|
|
|
85
|
+
| [Variance per Feature](https://scikit-learn.org/stable/modules/generated/sklearn.feature_selection.VarianceThreshold.html) | `threshold` |
|
|
86
|
+
| [Correlation pairwise Features](https://pandas.pydata.org/pandas-docs/stable/reference/api/pandas.DataFrame.corr.html) | [Pearson Correlation Coefficient](https://en.wikipedia.org/wiki/Pearson_correlation_coefficient) <br> [Kendall Rank Correlation Coefficient](https://en.wikipedia.org/wiki/Kendall_rank_correlation_coefficient) <br> [Spearman's Rank Correlation Coefficient](https://en.wikipedia.org/wiki/Spearman%27s_rank_correlation_coefficient) <br> |
|
|
87
|
+
| [Statistical Analysis](https://scikit-learn.org/stable/modules/feature_selection.html#univariate-feature-selection) | [ANOVA F-test Classification](https://scikit-learn.org/stable/modules/generated/sklearn.feature_selection.f_classif.html) <br> [F-value Regression](https://scikit-learn.org/stable/modules/generated/sklearn.feature_selection.f_regression.html) <br> [Chi-Square](https://scikit-learn.org/stable/modules/generated/sklearn.feature_selection.chi2.html) <br> [KL Divergence](https://en.wikipedia.org/wiki/Kullback–Leibler_divergence) <br> [Mutual Information Classification](https://scikit-learn.org/stable/modules/generated/sklearn.feature_selection.mutual_info_classif.html) <br> [Variance Inflation Factor](https://www.statsmodels.org/stable/generated/statsmodels.stats.outliers_influence.variance_inflation_factor.html) |
|
|
88
|
+
| [Linear Methods](https://en.wikipedia.org/wiki/Linear_regression) | [Linear Regression](https://scikit-learn.org/stable/modules/generated/sklearn.linear_model.LinearRegression.html?highlight=linear%20regression#sklearn.linear_model.LinearRegression) <br> [Logistic Regression](https://scikit-learn.org/stable/modules/generated/sklearn.linear_model.LogisticRegression.html?highlight=logistic%20regression#sklearn.linear_model.LogisticRegression) <br> [Lasso Regularization](https://scikit-learn.org/stable/modules/generated/sklearn.linear_model.Lasso.html#sklearn.linear_model.Lasso) <br> [Ridge Regularization](https://scikit-learn.org/stable/modules/generated/sklearn.linear_model.Ridge.html#sklearn.linear_model.Ridge) <br> |
|
|
89
|
+
| [Tree-based Methods](https://scikit-learn.org/stable/modules/tree.html) | [Decision Tree](https://scikit-learn.org/stable/modules/generated/sklearn.tree.DecisionTreeClassifier.html#sklearn.tree.DecisionTreeClassifier) <br> [Random Forest](https://scikit-learn.org/stable/modules/generated/sklearn.ensemble.RandomForestClassifier.html?highlight=random%20forest#sklearn.ensemble.RandomForestClassifier) <br> [Extra Trees Classifier](https://scikit-learn.org/stable/modules/generated/sklearn.ensemble.ExtraTreesClassifier.html) <br> [XGBoost](https://xgboost.readthedocs.io/en/latest/) <br> [LightGBM](https://lightgbm.readthedocs.io/en/latest/) <br> [AdaBoost](https://scikit-learn.org/stable/modules/generated/sklearn.ensemble.AdaBoostClassifier.html) <br> [CatBoost](https://github.com/catboost)<br> [Gradient Boosting Tree](http://scikit-learn.org/stable/modules/generated/sklearn.ensemble.GradientBoostingClassifier.html) <br> |
|
|
90
|
+
| [Text-based Methods](https://link.springer.com/chapter/10.1007/978-3-030-78230-6_27) | `featurization_method` = [TextWiser](https://github.com/fidelity/textwiser) <br> `optimization_method = ["exact", "greedy", "kmeans", "random"]` <br> `cost_metric = ["unicost", "diverse"]` |
|
|
91
|
+
|
|
92
|
+
|
|
93
|
+
|
|
94
|
+
## Benchmarking
|
|
95
|
+
|
|
96
|
+
```python
|
|
97
|
+
# Imports
|
|
98
|
+
from sklearn.datasets import fetch_california_housing
|
|
99
|
+
from feature.utils import get_data_label
|
|
100
|
+
from xgboost import XGBClassifier, XGBRegressor
|
|
101
|
+
from feature.selector import SelectionMethod, benchmark, calculate_statistics
|
|
102
|
+
|
|
103
|
+
# Data
|
|
104
|
+
data, label = get_data_label(fetch_california_housing())
|
|
105
|
+
|
|
106
|
+
# Selectors
|
|
107
|
+
corr_threshold = 0.5
|
|
108
|
+
num_features = 3
|
|
109
|
+
tree_params = {"n_estimators": 50, "max_depth": 5, "random_state": 111, "n_jobs": 4}
|
|
110
|
+
selectors = {
|
|
111
|
+
|
|
112
|
+
# Correlation methods
|
|
113
|
+
"corr_pearson": SelectionMethod.Correlation(corr_threshold, method="pearson"),
|
|
114
|
+
"corr_kendall": SelectionMethod.Correlation(corr_threshold, method="kendall"),
|
|
115
|
+
"corr_spearman": SelectionMethod.Correlation(corr_threshold, method="spearman"),
|
|
116
|
+
|
|
117
|
+
# Statistical methods
|
|
118
|
+
"stat_anova": SelectionMethod.Statistical(num_features, method="anova"),
|
|
119
|
+
"stat_chi_square": SelectionMethod.Statistical(num_features, method="chi_square"),
|
|
120
|
+
"stat_kl_divergence": SelectionMethod.Statistical(num_features, method="kl_divergence"),
|
|
121
|
+
"stat_mutual_info": SelectionMethod.Statistical(num_features, method="mutual_info"),
|
|
122
|
+
|
|
123
|
+
# Linear methods
|
|
124
|
+
"linear": SelectionMethod.Linear(num_features, regularization="none"),
|
|
125
|
+
"lasso": SelectionMethod.Linear(num_features, regularization="lasso", alpha=1000),
|
|
126
|
+
"ridge": SelectionMethod.Linear(num_features, regularization="ridge", alpha=1000),
|
|
127
|
+
|
|
128
|
+
# Non-linear tree-based methods
|
|
129
|
+
"random_forest": SelectionMethod.TreeBased(num_features),
|
|
130
|
+
"xgboost_classif": SelectionMethod.TreeBased(num_features, estimator=XGBClassifier(**tree_params)),
|
|
131
|
+
"xgboost_regress": SelectionMethod.TreeBased(num_features, estimator=XGBRegressor(**tree_params))
|
|
132
|
+
}
|
|
133
|
+
|
|
134
|
+
# Benchmark (sequential)
|
|
135
|
+
score_df, selected_df, runtime_df = benchmark(selectors, data, label, cv=5)
|
|
136
|
+
print(score_df, "\n\n", selected_df, "\n\n", runtime_df)
|
|
137
|
+
|
|
138
|
+
# Benchmark (in parallel)
|
|
139
|
+
score_df, selected_df, runtime_df = benchmark(selectors, data, label, cv=5, n_jobs=4)
|
|
140
|
+
print(score_df, "\n\n", selected_df, "\n\n", runtime_df)
|
|
141
|
+
|
|
142
|
+
# Get benchmark statistics by feature
|
|
143
|
+
stats_df = calculate_statistics(score_df, selected_df)
|
|
144
|
+
print(stats_df)
|
|
145
|
+
```
|
|
146
|
+
|
|
147
|
+
## Text-based Selection
|
|
148
|
+
This example shows how to use text-based selection. In this scenario, we would like to select a subset of articles that is most diverse in the text embedding space and covers a range of topics.
|
|
149
|
+
|
|
150
|
+
```python
|
|
151
|
+
# Import Selective and TextWiser
|
|
152
|
+
import pandas as pd
|
|
153
|
+
from feature.selector import Selective, SelectionMethod
|
|
154
|
+
from textwiser import TextWiser, Embedding, Transformation
|
|
155
|
+
|
|
156
|
+
# Data with the text content of each article
|
|
157
|
+
data = pd.DataFrame({"article_1": ["article text here"],
|
|
158
|
+
"article_2": ["article text here"],
|
|
159
|
+
"article_3": ["article text here"],
|
|
160
|
+
"article_4": ["article text here"],
|
|
161
|
+
"article_5": ["article text here"]})
|
|
162
|
+
|
|
163
|
+
# Labels to denote 0/1 coverage metadata for each article
|
|
164
|
+
# across four labels, e.g., sports, international, entertainment, science
|
|
165
|
+
labels = pd.DataFrame({"article_1": [1, 1, 0, 1],
|
|
166
|
+
"article_2": [0, 1, 0, 0],
|
|
167
|
+
"article_3": [0, 0, 1, 0],
|
|
168
|
+
"article_4": [0, 0, 1, 1],
|
|
169
|
+
"article_5": [1, 1, 1, 0]},
|
|
170
|
+
index=["label_1", "label_2", "label_3", "label_4"])
|
|
171
|
+
|
|
172
|
+
# TextWiser featurization method to create text embeddings
|
|
173
|
+
textwiser = TextWiser(Embedding.TfIdf(), Transformation.NMF(n_components=20))
|
|
174
|
+
|
|
175
|
+
# Text-based selection
|
|
176
|
+
# The goal is to select a subset of articles
|
|
177
|
+
# that is most diverse in the text embedding space of articles
|
|
178
|
+
# and covers the most labels in each topic
|
|
179
|
+
selector = Selective(SelectionMethod.TextBased(num_features=2, featurization_method=textwiser))
|
|
180
|
+
|
|
181
|
+
# Feature reduction
|
|
182
|
+
subset = selector.fit_transform(data, labels)
|
|
183
|
+
print("Reduction:", list(subset.columns))
|
|
184
|
+
```
|
|
185
|
+
|
|
186
|
+
## Visualization
|
|
187
|
+
|
|
188
|
+
```python
|
|
189
|
+
import pandas as pd
|
|
190
|
+
from sklearn.datasets import fetch_california_housing
|
|
191
|
+
from feature.utils import get_data_label
|
|
192
|
+
from feature.selector import SelectionMethod, Selective, plot_importance
|
|
193
|
+
|
|
194
|
+
# Data
|
|
195
|
+
data, label = get_data_label(fetch_california_housing())
|
|
196
|
+
|
|
197
|
+
# Feature Selector
|
|
198
|
+
selector = Selective(SelectionMethod.Linear(num_features=8, regularization="none"))
|
|
199
|
+
subset = selector.fit_transform(data, label)
|
|
200
|
+
|
|
201
|
+
# Plot Feature Importance
|
|
202
|
+
df = pd.DataFrame(selector.get_absolute_scores(), index=data.columns)
|
|
203
|
+
plot_importance(df)
|
|
204
|
+
```
|
|
205
|
+
|
|
206
|
+
## Installation
|
|
207
|
+
|
|
208
|
+
Selective requires **Python 3.8+** and can be installed from PyPI using ``pip install selective``.
|
|
209
|
+
|
|
210
|
+
## Source
|
|
211
|
+
|
|
212
|
+
Alternatively, you can build a wheel package on your platform from scratch using the source code:
|
|
213
|
+
|
|
214
|
+
```bash
|
|
215
|
+
git clone https://github.com/fidelity/selective.git
|
|
216
|
+
cd selective
|
|
217
|
+
pip install setuptools wheel # if wheel is not installed
|
|
218
|
+
python setup.py sdist bdist_wheel
|
|
219
|
+
pip install dist/selective-X.X.X-py3-none-any.whl
|
|
220
|
+
```
|
|
221
|
+
|
|
222
|
+
## Test your setup
|
|
223
|
+
|
|
224
|
+
```
|
|
225
|
+
cd selective
|
|
226
|
+
python -m unittest discover tests
|
|
227
|
+
```
|
|
228
|
+
|
|
229
|
+
## Citation
|
|
230
|
+
|
|
231
|
+
If you use Selective in a publication, please cite it as:
|
|
232
|
+
|
|
233
|
+
```bibtex
|
|
234
|
+
@article{DBLP:journals/amai/HaDVH98,
|
|
235
|
+
author = {Kad\i{}o\u{g}lu, Serdar and Kleynhans, Bernard and Wang, Xin},
|
|
236
|
+
title = {Integrating optimized item selection with active learning for continuous exploration in recommender systems},
|
|
237
|
+
journal = {Ann. Math. Artif. Intell.},
|
|
238
|
+
year = {2024},
|
|
239
|
+
url = {https://doi.org/10.1007/s10472-024-09941-x},
|
|
240
|
+
doi = {10.1007/s10472-024-09941-x},
|
|
241
|
+
}
|
|
242
|
+
}
|
|
243
|
+
```
|
|
244
|
+
|
|
245
|
+
## Support
|
|
246
|
+
|
|
247
|
+
Please submit bug reports and feature requests as [Issues](https://github.com/fidelity/selective/issues).
|
|
248
|
+
|
|
249
|
+
## License
|
|
250
|
+
Selective is licensed under [Apache 2.0](https://github.com/fidelity/selective/blob/master/LICENSE.md)
|
|
@@ -4,9 +4,6 @@
|
|
|
4
4
|
# Selective: Feature Selection Library
|
|
5
5
|
**Selective** is a white-box feature selection library that supports supervised and unsupervised selection methods for classification and regression tasks.
|
|
6
6
|
|
|
7
|
-
Selective also provides optimized item selection based on diversity of text embeddings (via [TextWiser](https://github.com/fidelity/textwiser)) and
|
|
8
|
-
the coverage of binary labels by solving a multi-objective optimization problem ([CPAIOR'21](https://link.springer.com/chapter/10.1007/978-3-030-78230-6_27), [DSO@IJCAI'22](https://arxiv.org/abs/2112.03105)). The approach showed to speed-up online experimentation significantly and boost recommender systems [NVIDIA GTC'22](https://www.youtube.com/watch?v=_v-B2nRy79w).
|
|
9
|
-
|
|
10
7
|
The library provides:
|
|
11
8
|
|
|
12
9
|
* Simple to complex selection methods: Variance, Correlation, Statistical, Linear, Tree-based, or Customized.
|
|
@@ -16,6 +13,9 @@ The library provides:
|
|
|
16
13
|
* Benchmarking multiple selectors using cross-validation with built-in parallelization.
|
|
17
14
|
* Inspection of the results and feature importance.
|
|
18
15
|
|
|
16
|
+
Selective also provides optimized item selection based on diversity of text embeddings via [TextWiser](https://github.com/fidelity/textwiser) and
|
|
17
|
+
coverage of binary labels via multi-objective optimization ([AMAI'24](https://trebuchet.public.springernature.app/get_content/2c9eb6df-5c2b-42bc-89d6-4e3eb8bc8799?utm_source=rct_congratemailt&utm_medium=email&utm_campaign=nonoa_20240405&utm_content=10.1007/s10472-024-09941-x), [CPAIOR'21](https://link.springer.com/chapter/10.1007/978-3-030-78230-6_27), [DSO@IJCAI'22](https://arxiv.org/abs/2112.03105)). This approach speeds-up online experimentation and boosts recommender systems significantly as presented at [NVIDIA GTC'22](https://www.youtube.com/watch?v=_v-B2nRy79w).
|
|
18
|
+
|
|
19
19
|
Selective is developed by the Artificial Intelligence Center of Excellence at Fidelity Investments.
|
|
20
20
|
|
|
21
21
|
## Quick Start
|
|
@@ -48,7 +48,7 @@ print("Scores:", list(selector.get_absolute_scores()))
|
|
|
48
48
|
|:--------------------------------------------------------------------------------------------------------------------------:|:--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------:|
|
|
49
49
|
| [Variance per Feature](https://scikit-learn.org/stable/modules/generated/sklearn.feature_selection.VarianceThreshold.html) | `threshold` |
|
|
50
50
|
| [Correlation pairwise Features](https://pandas.pydata.org/pandas-docs/stable/reference/api/pandas.DataFrame.corr.html) | [Pearson Correlation Coefficient](https://en.wikipedia.org/wiki/Pearson_correlation_coefficient) <br> [Kendall Rank Correlation Coefficient](https://en.wikipedia.org/wiki/Kendall_rank_correlation_coefficient) <br> [Spearman's Rank Correlation Coefficient](https://en.wikipedia.org/wiki/Spearman%27s_rank_correlation_coefficient) <br> |
|
|
51
|
-
| [Statistical Analysis](https://scikit-learn.org/stable/modules/feature_selection.html#univariate-feature-selection) | [ANOVA F-test Classification](https://scikit-learn.org/stable/modules/generated/sklearn.feature_selection.f_classif.html) <br> [F-value Regression](https://scikit-learn.org/stable/modules/generated/sklearn.feature_selection.f_regression.html) <br> [Chi-Square](https://scikit-learn.org/stable/modules/generated/sklearn.feature_selection.chi2.html) <br> [Mutual Information Classification](https://scikit-learn.org/stable/modules/generated/sklearn.feature_selection.mutual_info_classif.html) <br> [Variance Inflation Factor](https://www.statsmodels.org/stable/generated/statsmodels.stats.outliers_influence.variance_inflation_factor.html)
|
|
51
|
+
| [Statistical Analysis](https://scikit-learn.org/stable/modules/feature_selection.html#univariate-feature-selection) | [ANOVA F-test Classification](https://scikit-learn.org/stable/modules/generated/sklearn.feature_selection.f_classif.html) <br> [F-value Regression](https://scikit-learn.org/stable/modules/generated/sklearn.feature_selection.f_regression.html) <br> [Chi-Square](https://scikit-learn.org/stable/modules/generated/sklearn.feature_selection.chi2.html) <br> [KL Divergence](https://en.wikipedia.org/wiki/Kullback–Leibler_divergence) <br> [Mutual Information Classification](https://scikit-learn.org/stable/modules/generated/sklearn.feature_selection.mutual_info_classif.html) <br> [Variance Inflation Factor](https://www.statsmodels.org/stable/generated/statsmodels.stats.outliers_influence.variance_inflation_factor.html) |
|
|
52
52
|
| [Linear Methods](https://en.wikipedia.org/wiki/Linear_regression) | [Linear Regression](https://scikit-learn.org/stable/modules/generated/sklearn.linear_model.LinearRegression.html?highlight=linear%20regression#sklearn.linear_model.LinearRegression) <br> [Logistic Regression](https://scikit-learn.org/stable/modules/generated/sklearn.linear_model.LogisticRegression.html?highlight=logistic%20regression#sklearn.linear_model.LogisticRegression) <br> [Lasso Regularization](https://scikit-learn.org/stable/modules/generated/sklearn.linear_model.Lasso.html#sklearn.linear_model.Lasso) <br> [Ridge Regularization](https://scikit-learn.org/stable/modules/generated/sklearn.linear_model.Ridge.html#sklearn.linear_model.Ridge) <br> |
|
|
53
53
|
| [Tree-based Methods](https://scikit-learn.org/stable/modules/tree.html) | [Decision Tree](https://scikit-learn.org/stable/modules/generated/sklearn.tree.DecisionTreeClassifier.html#sklearn.tree.DecisionTreeClassifier) <br> [Random Forest](https://scikit-learn.org/stable/modules/generated/sklearn.ensemble.RandomForestClassifier.html?highlight=random%20forest#sklearn.ensemble.RandomForestClassifier) <br> [Extra Trees Classifier](https://scikit-learn.org/stable/modules/generated/sklearn.ensemble.ExtraTreesClassifier.html) <br> [XGBoost](https://xgboost.readthedocs.io/en/latest/) <br> [LightGBM](https://lightgbm.readthedocs.io/en/latest/) <br> [AdaBoost](https://scikit-learn.org/stable/modules/generated/sklearn.ensemble.AdaBoostClassifier.html) <br> [CatBoost](https://github.com/catboost)<br> [Gradient Boosting Tree](http://scikit-learn.org/stable/modules/generated/sklearn.ensemble.GradientBoostingClassifier.html) <br> |
|
|
54
54
|
| [Text-based Methods](https://link.springer.com/chapter/10.1007/978-3-030-78230-6_27) | `featurization_method` = [TextWiser](https://github.com/fidelity/textwiser) <br> `optimization_method = ["exact", "greedy", "kmeans", "random"]` <br> `cost_metric = ["unicost", "diverse"]` |
|
|
@@ -81,6 +81,7 @@ selectors = {
|
|
|
81
81
|
# Statistical methods
|
|
82
82
|
"stat_anova": SelectionMethod.Statistical(num_features, method="anova"),
|
|
83
83
|
"stat_chi_square": SelectionMethod.Statistical(num_features, method="chi_square"),
|
|
84
|
+
"stat_kl_divergence": SelectionMethod.Statistical(num_features, method="kl_divergence"),
|
|
84
85
|
"stat_mutual_info": SelectionMethod.Statistical(num_features, method="mutual_info"),
|
|
85
86
|
|
|
86
87
|
# Linear methods
|
|
@@ -168,7 +169,7 @@ plot_importance(df)
|
|
|
168
169
|
|
|
169
170
|
## Installation
|
|
170
171
|
|
|
171
|
-
Selective requires **Python 3.
|
|
172
|
+
Selective requires **Python 3.8+** and can be installed from PyPI using ``pip install selective``.
|
|
172
173
|
|
|
173
174
|
## Source
|
|
174
175
|
|
|
@@ -189,9 +190,25 @@ cd selective
|
|
|
189
190
|
python -m unittest discover tests
|
|
190
191
|
```
|
|
191
192
|
|
|
193
|
+
## Citation
|
|
194
|
+
|
|
195
|
+
If you use Selective in a publication, please cite it as:
|
|
196
|
+
|
|
197
|
+
```bibtex
|
|
198
|
+
@article{DBLP:journals/amai/HaDVH98,
|
|
199
|
+
author = {Kad\i{}o\u{g}lu, Serdar and Kleynhans, Bernard and Wang, Xin},
|
|
200
|
+
title = {Integrating optimized item selection with active learning for continuous exploration in recommender systems},
|
|
201
|
+
journal = {Ann. Math. Artif. Intell.},
|
|
202
|
+
year = {2024},
|
|
203
|
+
url = {https://doi.org/10.1007/s10472-024-09941-x},
|
|
204
|
+
doi = {10.1007/s10472-024-09941-x},
|
|
205
|
+
}
|
|
206
|
+
}
|
|
207
|
+
```
|
|
208
|
+
|
|
192
209
|
## Support
|
|
193
210
|
|
|
194
211
|
Please submit bug reports and feature requests as [Issues](https://github.com/fidelity/selective/issues).
|
|
195
212
|
|
|
196
213
|
## License
|
|
197
|
-
Selective is licensed under
|
|
214
|
+
Selective is licensed under [Apache 2.0](https://github.com/fidelity/selective/blob/master/LICENSE.md)
|
|
@@ -0,0 +1,66 @@
|
|
|
1
|
+
# -*- coding: utf-8 -*-
|
|
2
|
+
# Copyright FMR LLC <opensource@fidelity.com>
|
|
3
|
+
# SPDX-License-Identifier: Apache-2.0
|
|
4
|
+
|
|
5
|
+
from typing import NoReturn, Tuple
|
|
6
|
+
|
|
7
|
+
import pandas as pd
|
|
8
|
+
import numpy as np
|
|
9
|
+
|
|
10
|
+
from scipy.special import rel_entr
|
|
11
|
+
from feature.base import _BaseSupervisedSelector
|
|
12
|
+
from feature.utils import Num, check_true
|
|
13
|
+
|
|
14
|
+
|
|
15
|
+
class _KL_Divergence(_BaseSupervisedSelector):
|
|
16
|
+
|
|
17
|
+
def __init__(self, seed: int, num_features: Num, num_bins: Num = 100):
|
|
18
|
+
super().__init__(seed)
|
|
19
|
+
|
|
20
|
+
self.num_features = num_features # this could be int or float
|
|
21
|
+
self.num_bins = num_bins
|
|
22
|
+
|
|
23
|
+
def fit(self, X: pd.DataFrame, y: pd.Series) -> NoReturn:
|
|
24
|
+
|
|
25
|
+
label_categories = np.unique(y)
|
|
26
|
+
check_true(len(label_categories) == 2, TypeError("Only binary labels are supported for KL Divergence"))
|
|
27
|
+
input_dimension = X.shape[1]
|
|
28
|
+
|
|
29
|
+
kl_mat = np.zeros((input_dimension, 1))
|
|
30
|
+
X = X.values
|
|
31
|
+
|
|
32
|
+
class_one_idx = np.where(y == label_categories[0])[0]
|
|
33
|
+
class_two_idx = np.where(y == label_categories[1])[0]
|
|
34
|
+
|
|
35
|
+
for i in range(input_dimension):
|
|
36
|
+
|
|
37
|
+
# Create two distributions, one for the positive label and one for the negative label
|
|
38
|
+
f1 = np.histogram(X[class_one_idx, i], bins = self.num_bins)[0]
|
|
39
|
+
f2 = np.histogram(X[class_two_idx, i], bins = self.num_bins)[0]
|
|
40
|
+
|
|
41
|
+
# Normalize the distributions to be between 0 and 1
|
|
42
|
+
f1 = f1 / np.sum(f1)
|
|
43
|
+
f2 = f2 / np.sum(f2)
|
|
44
|
+
|
|
45
|
+
# KL Divergence is not symmetric, so we calculate divergence in both directions
|
|
46
|
+
kl = rel_entr(f1, f2)
|
|
47
|
+
kl_reversed = rel_entr(f2, f1)
|
|
48
|
+
|
|
49
|
+
# The relative entropy function returns KL(P || Q) = np.inf when P == 0 and Q != 0.
|
|
50
|
+
kl[kl == np.inf] = 9999
|
|
51
|
+
kl_reversed[kl_reversed == np.inf] = 9999
|
|
52
|
+
|
|
53
|
+
# The final score is the combination of KL divergence in both directions.
|
|
54
|
+
# This could possibly be a flag in a future version to determine which direction to apply KL Divergence
|
|
55
|
+
# in if bidirectional is not desired.
|
|
56
|
+
kl_mat[i] = np.sum(kl) + np.sum(kl_reversed)
|
|
57
|
+
|
|
58
|
+
scores_ = kl_mat.flatten()
|
|
59
|
+
|
|
60
|
+
self.scores_ = scores_ # This is used by the statistical.py fit function.
|
|
61
|
+
self.abs_scores = scores_
|
|
62
|
+
|
|
63
|
+
def transform(self, data: pd.DataFrame) -> pd.DataFrame:
|
|
64
|
+
|
|
65
|
+
# Select top-k from data based on abs_scores and num_features
|
|
66
|
+
return self.get_top_k(data, self.abs_scores)
|