6ce198c78f
[MRG] #6581 n_samples of utils.resample can be more when replace is True
...
* #6581 n_samples can be more when replace is True
* more compact code
2016-04-02 08:58:40 +02:00
e34bbc1002
Use fused type in inplace normalize
2016-03-26 10:38:17 +08:00
42d49c8c14
Make assign_rows_csr support Cython fused types
2016-03-25 00:25:36 +08:00
328ebfa865
ENH RandomizedPCA collapsed into PCA
2016-03-11 09:10:37 +11:00
1784bf3f26
Merge pull request #5863 from trevorstephens/cw_util_fix
...
[MRG] Fix for missing classes found in y - Fixes #4327
2016-02-11 11:42:47 +01:00
cfa498ca68
FIX in randomized_svd flip sign
...
Flip sign according to `u` in both cases of `transpose`.
2016-02-08 16:59:44 +01:00
6eca250347
MAINT speed up test_extmath
2016-02-06 18:55:23 +01:00
0485ada58b
General spelling fixes
2015-12-16 09:46:42 -08:00
79006fa1fe
Provide 'self' argument to instance method
2015-12-12 01:46:56 -05:00
9f136ff293
ENH add multinomial SAG solver for LogisticRegression
2015-12-04 18:36:37 +01:00
2738d59ab4
fix for missing classes found in y #4327
2015-11-16 16:36:01 -08:00
776e53b127
skip unstable tests on 32bit platform
2015-11-02 15:21:22 -05:00
8ff339eb5b
Update test_utils.py
...
Comment instead of docstring to follow sklearn convention.
Regression test made easier by adding small value to main diagonal of s.p.d. matrix
2015-10-23 08:57:31 +02:00
6cb51b2a28
Initialize ARPACK eigsh
...
`v0 = random_state.rand(M.shape[0])` leads to an initial residual vector in ARPACK which is all positive. However, this is not the absolute or squared residual, but a true difference. Thus, it is better to initialize with `v0=random_state.uniform(-1, 1, M.shape[0])` to have an equally distributed sign. This is the way that ARPACK initializes the residuals.
The effect of the previous initialization is that eigsh frequently does not converge to the correct eigenvalues, e.g. negative eigenvalues for s.p.d. matrix, which leads to an incorrect null-space.
- initialized all occurences of sklearn.utils.arpack.eigsh the same way it would be initialzed by ARPACK
- regression test to test behavior of new initialization
2015-10-23 08:57:30 +02:00
744d161d37
Merge pull request #5431 from hlin117/nan-targets
...
[MRG + 2] Add check to regression models to raise error when targets are NaN
2015-10-22 09:36:12 +02:00
b18f2951e4
randomized_svd: power iter, normalization, benchmark
2015-10-21 11:21:40 +02:00
c657d77c3c
#5322 : Regression test for test_check_estimator
2015-10-20 10:36:48 -05:00
7f820269b8
#5322 : Fixing test in test_estimator_checks.test_check_estimator
2015-10-20 10:36:47 -05:00
857cb09254
FIX/ENH Make the moved class resilient to the deprecation patching
2015-10-20 01:24:35 +02:00
a87011262a
FIX precision to float64 across the codebase
2015-10-16 11:20:29 +02:00
85223b9bcc
a common test to check if classifiers fail when fed regression targets
2015-10-14 14:46:13 -04:00
6a5a2f7960
partial_fit for scalers
2015-10-13 11:37:00 +02:00
c805fbc79b
Refactor tests
2015-10-10 22:05:40 -04:00
0bc3a6fbca
Merge pull request #5245 from ogrisel/lda-acronym-deprecation
...
[MRG+1] Deprecate LDA/QDA in favor of expanded names
2015-09-14 13:35:06 +02:00
8424c48d68
Merge pull request #5262 from andylamb/lamb-fix-class-weight-check
...
[MRG + 1] Fix check in `compute_class_weight`.
2015-09-14 00:09:22 -04:00
9ec03c276f
Change formatting of `assert_raise_message`.
2015-09-13 09:16:25 -07:00
f01a695ca8
Fix assertion for Python 3, use `assert_raise_message`.
2015-09-12 11:48:41 -07:00
581056774a
Fix check in `compute_class_weight`.
...
In https://github.com/scikit-learn/scikit-learn/issues/4921 , if a key in `class_weights`, a user-specified dictionary of classes, is not present in `classes`, an error should be raised.
Added unittest.
2015-09-12 10:19:24 -07:00
679936f6bc
MAINT Deprecate LDA/QDA in favor of expanded names
...
The LDA accronym for Linear Discriminant Analysis is ambiguous
because of the newly introduced Latent Dirichlet Allocation model.
We therefore deprecate the sklearn.lda.LDA and sklearn.lda.QDA
in favor of explicit names.
2015-09-11 15:32:11 +02:00
94eb61960a
ENH add sag solver in LogisticRegression and Ridge
2015-09-10 13:28:02 -04:00
cb591e1d3e
Merge pull request #5225 from MechCoder/fix_overflow
...
[MRG + 1] Add numerically stable softmax function to utils.extmath
2015-09-09 11:47:35 -04:00
2f099334f9
MAINT deprecate 1d input arrays for all estimators
...
Passing 1D arrays to check_array, without setting `ensure_2d` to false now
raises a deprecation warning before reshaping it. This will later throw an
error.
All Scaler classes also throw warnings when 1D arrays are passed.
All unit tests/doctests are modified to ensure that no 1D arrays are passed,
except in explicit 1D array tests where the warnings have been silenced.
Additional tests are also included which check for different 1D array cases.
2D array tests with one samples and one features are also added and where
they failed, `check_array` call has been modified to give a more useful error
message
2015-09-09 15:49:58 +02:00
d81b8ed898
Add test for softmax function
2015-09-08 17:34:33 -04:00
ef66046c12
Merge pull request #5094 from pv/fix-inplace
...
[MRG] TST: fix undefined behavior in test
2015-08-23 20:48:23 +10:00
18221db9fd
FIX type_of_target will now raise ValueError on ml-seq; [[1, 2]] is mc-mo;
2015-08-11 22:27:15 +05:30
c031b39787
MAINT is_label_indicator_matrix --> is_multilabel
2015-08-11 22:27:15 +05:30
7def0a75cd
TST/MAINT reintroduce mixed types test sans seq. of seq.
2015-08-11 22:27:15 +05:30
8f1ad4437f
FIX 2D binary array-like must be considered as unknown
2015-08-11 22:27:14 +05:30
72f6225f7b
MAINT/ENH inline the logic for _is_sequence_of_sequences
2015-08-11 22:27:14 +05:30
664d78eb7c
MAINT Remove support for the deprecated sequence of sequences
...
MAINT Remove sequence of sequence support from datasets
MAINT Remove return_indicator param
MAINT Remove multilabel-seq test in OVR
MAINT Remove multilable-seq test in check_cv
MAINT Remove multilabel seq test in label_binarizer
TST type_of_target returns "unknown" for multilabel-sequence types
TST _check_targets should raise a ValueError
DOC show multilabel indicator as an example; remove return_indicator param
DOC use consistent lower case y for target
2015-08-11 22:26:44 +05:30
d04d7cffd4
TST: fix undefined behavior in test
2015-08-06 23:44:14 +03:00
cec3bf98f2
FIX assign_rows_csr: should not zero the entire out array
...
Fixes #4985 .
2015-07-31 10:59:23 +02:00
01aff46402
Merge pull request #4347 from amueller/class_weight_auto
...
[MRG+1] Use more natural class_weight="auto" heuristic
2015-06-01 16:15:37 -04:00
2ddb503b59
Cosmit readability improvements and better whatsnew.
2015-06-01 12:56:21 -04:00
eedc1cddab
Use more natural class_weight="auto" heuristic
2015-06-01 12:49:53 -04:00
a8626b36a6
TST/COSMIT remove nose call boilerplate
2015-05-28 14:54:01 -04:00
0ee4d973b5
Merge pull request #4751 from amueller/dict_learning_n_jobs_bug
...
[MRG+1] FIX n_jobs slicing bug in dict learning
2015-05-22 13:55:18 +02:00
5cc234a849
Merge pull request #4716 from TomDLT/notfitted
...
[MRG + 1] check that unfitted estimators raise ValueError
2015-05-21 11:10:42 -04:00
5029db7f32
Merge pull request #4679 from TomDLT/newton_cg
...
[MRG+1] change func_grad_hess into grad_hess
2015-05-21 09:07:45 +02:00
990a6b402f
FIX n_jobs slicing bug in dict learning, add input validation to gen_even_slices
2015-05-20 15:29:25 -04:00
ningchi