Lars Buitinck
|
b77804dcb4
|
BUG NaN handling in SelectPercentile and SelectKBest
Including non-regression test.
|
2012-10-16 12:53:27 +02:00 |
Lars Buitinck
|
19a1a41379
|
ENH perform feature selection on scores, not p-values, when possible
More stable in the case of chi2; allows for feature selectors that do not
perform significance tests.
|
2012-10-04 12:13:50 +02:00 |
Lars Buitinck
|
aa69c564d7
|
COSMIT error messages in GenericUnivariateSelect
|
2012-10-04 12:11:48 +02:00 |
Lars Buitinck
|
2e48248531
|
COSMIT + DOC univariate feature selection
rm deprecated properties
input validation in SelectPercentile
use super()
docstring fixes
|
2012-09-18 14:38:52 +02:00 |
Andreas Mueller
|
1373bcd7b7
|
DOC document attributes scores_ and pvalues_ in feature selection docstrings, some superficial cleanup.
|
2012-09-17 21:50:24 +01:00 |
Andreas Mueller
|
66c8a706d2
|
ENH fix percentile tiebreaking, add warning
|
2012-09-17 21:40:58 +01:00 |
Lars Buitinck
|
b4aa0cbca1
|
DOC RFE docstring + link RFECV in narrative docs
|
2012-09-17 20:16:38 +02:00 |
Lars Buitinck
|
ae24236681
|
BUG allow array-like y in RFE
+ cosmit: use np.ravel to flatten np.matrix
|
2012-09-17 19:37:19 +02:00 |
Andreas Mueller
|
ea3f32ba7a
|
FIX typo
|
2012-09-12 20:14:09 +01:00 |
Andreas Mueller
|
aa204678ee
|
MISC rfe outputs loss, not score
|
2012-09-12 20:11:23 +01:00 |
Andreas Mueller
|
6fff2657e8
|
ENH fixed bug in sparse RFECV, fixed bug in RFECV init (step was always 1), added decision_function and predict_proba to RFE and RFECV
|
2012-09-12 20:09:59 +01:00 |
Andreas Mueller
|
57574dc031
|
ENH issue 1115: grid search support for rfe via ``estimator_params``
|
2012-09-12 20:09:58 +01:00 |
Andreas Mueller
|
e14dd67d1b
|
ENH address issue 1028: clone estimator in RFE
|
2012-09-12 20:09:58 +01:00 |
Lars Buitinck
|
5a320d703b
|
COSMIT replace useless safe_sparse_dot in chi2 with np.dot
|
2012-09-08 23:08:07 +02:00 |
Vlad Niculae
|
7f76aa1def
|
Add __all__ for half of the scikit
|
2012-09-04 18:00:49 +01:00 |
Vlad Niculae
|
9e7c03f71a
|
Add or fix deprecation schedule in warnings.
|
2012-09-01 20:02:14 +02:00 |
Andreas Mueller
|
2ec38b31e1
|
ENH rudimentary testing of tranformer objects
|
2012-09-01 18:25:05 +02:00 |
Andreas Mueller
|
eb89d5a173
|
FIX Worst feature in RFECV missing. closes issue #681.
|
2012-08-26 18:34:15 +02:00 |
Andreas Mueller
|
bef749b4e8
|
ENH sparse matrix support for RFE and RFECV. Closes issue #1018.
|
2012-08-16 18:05:16 +01:00 |
Andreas Mueller
|
c789e3c201
|
ENH simplify as @mblondel suggests
|
2012-08-16 17:51:02 +01:00 |
Andreas Mueller
|
52b1bef4a5
|
ENH Addressed @mblondel's comments.
|
2012-08-16 17:51:01 +01:00 |
Andreas Mueller
|
1aec782c9c
|
COSMIT Fixed commata
|
2012-08-16 17:51:01 +01:00 |
Andreas Mueller
|
00aa08971d
|
DOC Polished some docstrings
|
2012-08-16 17:51:00 +01:00 |
Andreas Mueller
|
fc3cdc7472
|
TST Simplified tests, test that sparse and dense versions give the same result, always return arrays, not matrices.
|
2012-08-16 17:51:00 +01:00 |
Andreas Mueller
|
c520275f06
|
ENH sparse matrix support in univariate feature selection
|
2012-08-16 17:50:59 +01:00 |
Andreas Mueller
|
ba4f3edb97
|
DOC explained RFE default behavior in docstring.
|
2012-08-12 18:17:54 +01:00 |
Andreas Mueller
|
c41b03bd35
|
FIX feature selection dies on non-csr sparse matrices (that are unsubscribable). Regression test should go in common testing.
|
2012-08-11 16:11:49 +01:00 |
John Benediktsson
|
ce856db1db
|
COSMIT: fix more typos.
|
2012-08-09 15:50:42 -07:00 |
John Benediktsson
|
8b294867df
|
COSMIT: fix reference to scipy.stats.kruskal.
|
2012-08-09 15:49:26 -07:00 |
John Benediktsson
|
a733f3b6ff
|
COSMIT: fix typo of homoscedasticity.
|
2012-08-09 15:47:26 -07:00 |
John Benediktsson
|
9bf746cc50
|
COSMIT: doc fixes to sklearn.feature_selection.univariate_selection.
|
2012-08-09 15:41:05 -07:00 |
Mathieu Blondel
|
3fe58758cd
|
Credits and license.
|
2012-07-24 22:31:42 +09:00 |
Mathieu Blondel
|
80745240da
|
Test default input.
|
2012-07-24 22:31:29 +09:00 |
Lars Buitinck
|
4701319d9c
|
TST: improve coverage of feature_selection.SelectorMixin
|
2012-07-24 12:40:54 +02:00 |
Andreas Mueller
|
f71185dca9
|
ENH default score func for univariate feature selection: f_classif
|
2012-06-26 14:16:04 +02:00 |
Andreas Mueller
|
04d81608a1
|
ENH Add MetaEstimatorMixin, make RFE default constructible
|
2012-06-26 14:16:03 +02:00 |
Andreas Mueller
|
86cedafa08
|
ENH make all Estimators default constructible (except SparseCoder)
|
2012-06-26 14:16:03 +02:00 |
Andreas Mueller
|
1ebcc4a486
|
COSMIT pep8
|
2012-05-15 21:18:30 +02:00 |
Gael Varoquaux
|
065b905aed
|
BUG: casting error with recent numpys
|
2012-05-12 00:02:30 +02:00 |
Andreas Mueller
|
eeb6dcc4cc
|
ENH use random states everywhere, never call np.random.
|
2012-05-06 18:35:14 +02:00 |
Andreas Mueller
|
3e5f2e5e08
|
ENH backport "assert_less" and "assert_greater", rename "assert_lower" and use it everywhere :)
|
2012-05-06 15:43:29 +02:00 |
Andreas Mueller
|
77954f292e
|
FIXed many tests
|
2012-05-05 21:57:04 +02:00 |
Lars Buitinck
|
258f19383c
|
BUG ensure that SelectKBest actually selects k features
Fixes Issue #805
|
2012-04-30 17:25:13 +02:00 |
Lars Buitinck
|
b17a82023c
|
ENH export f_oneway from feature_selection module
+ import in test as a user would
|
2012-04-30 17:10:55 +02:00 |
Lars Buitinck
|
b22eb2ac9c
|
BUG chi² feature selection didn't work for COO matrices
Reported via StackOverflow, http://stackoverflow.com/q/10373317/166749
|
2012-04-30 14:32:32 +02:00 |
Lars Buitinck
|
0e8e38e3b2
|
DOC typo
|
2012-04-27 14:47:40 +02:00 |
Alexandre Gramfort
|
2d5c2a7fe0
|
FIX : array copy for compat pb
|
2012-03-29 16:11:08 +02:00 |
Alexandre Gramfort
|
eb9ad47594
|
ENH : speed up f_regression
|
2012-03-29 15:21:08 +02:00 |
Satrajit Ghosh
|
239039c84c
|
tst: added test for feature selection. this test would have failed in the previous case. closes #727
|
2012-03-26 09:59:07 -04:00 |
Satrajit Ghosh
|
efe14a7a98
|
fix: force copy to True in case underlying default behavior changes.
|
2012-03-26 09:54:25 -04:00 |