41cbfded7c
Add check for sample_weights
2015-10-23 00:06:07 -04:00
80e22b3ddf
Patch liblinear for sample_weights in LogisticRegression(and CV)
2015-10-23 00:00:31 -04:00
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
7b3da77ca0
#5322 : Removing unused ConvergenceWarning import in svm/base.py
2015-10-21 00:07:53 -05:00
52a6acc15c
#5322 : Addressing @MechCoder's changes
2015-10-21 00:04:42 -05:00
fffa98ad1f
MAINT Remove deprecated stuff that will be removed in 0.18
...
MAINT Remove support for gamma == 0.0
MAINT Remove support for uppercase values for loss and penalty
2015-10-20 18:24:35 +02:00
b1a1aa8395
#5322 : Resolving errors for SVR and NuSVR
2015-10-20 10:36:47 -05:00
e3afc0e8c9
MAINT move custom error/warning classes into sklearn.exceptions
...
ENH NonBLASDotWarning -> EfficiencyWarning; Improve error message
DOC Add exceptions module to modules/classes.rst
MAINT Move ConvergenceWarning, UndefinedMetricWarning et al into exceptions
MAINT Remove ChangedBehaviorWarning from base
DOC/FIX Improve DataConversionWarning's docstring
2015-10-19 22:35:35 +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
2afbc0f7e2
FIX make sure y is 1d in svm regressor, add common test for regressor shapes.
2015-07-31 10:34:39 -04:00
0e15b030d0
Add a space to an error message
2015-07-11 11:42:52 -05:00
5aa5a961bc
Correctly match penalty and loss with arguments
2015-07-11 11:42:11 -05:00
d517846cc5
FIX #4902 : SV*.predict_proba visible before fit
2015-07-06 12:35:47 +10:00
4c4bca1715
fix SVM can be tricked into running proba() #4791
2015-06-08 16:48:42 -04:00
567a956338
Merge branch 'pr/4714' + pep8 fixes
...
Conflicts:
doc/modules/model_persistence.rst
doc/modules/pipeline.rst
doc/modules/svm.rst
doc/tutorial/basic/tutorial.rst
doc/tutorial/statistical_inference/supervised_learning.rst
sklearn/svm/classes.py
2015-06-08 11:31:51 +02:00
41788c353b
Changed SVM to use gamma='auto' by default. Deprecated gamma=0.0.
2015-06-05 22:47:43 +01:00
6d4bcda876
ENH refactor OVO decision function, use it in SVC for sklearn-like decision_function shape
2015-06-05 13:14:45 -04:00
eedc1cddab
Use more natural class_weight="auto" heuristic
2015-06-01 12:49:53 -04:00
188fb11968
COSMIT use consistent shape description in docstring.
2015-04-01 10:24:58 -04:00
d89c215013
Add tags to classifiers and regressors to identify them as such.
2015-03-31 19:59:49 -04:00
b1ddbf4695
Fixes to sparse decision_function, rebase fixes.
2015-03-26 11:44:39 -04:00
0dc37f5f73
Adding sparse support for decision function
...
Adding test for decision_function on sparse matrices
Fixing typos
Fix Memory Leak
PEP8 fixes
COSMIT PEP8
Fixed class_weight_label problem
Fixed label problem in tests
renamed label_ to classes_
Fixed OneClass bug
Adding kernel tests for sparse matrices
Pass dense data to decision function
Don't check precomputed kernel
use safe_sparse_dot in test, regenerate cython
create a core-dump by testing the binary case... hurray?
2015-03-26 11:35:36 -04:00
119bc20abb
Merge pull request #4377 from vortex-ape/intercept_scaling
...
[MRG + 1] Fixes #4374 : LinearSVC(intercept_scaling=0) breaks
2015-03-20 12:56:11 +01:00
ea932afdb6
Added docstrings for kernel='precomputed'
...
removed spaces
2015-03-19 21:22:19 +05:30
a4f8f083a5
Specified original intercept_scaling value
2015-03-19 20:27:11 +05:30
5f44a8e2a3
Raising an error when intercept_scaling is zero
...
Added test for intercept_scaling
Fixed silly mistake
Added assert_raise_message
added test for logistic regression
removed None
added more tests
2015-03-17 23:23:39 +05:30
7a97192153
Minor fix for a comment
2015-03-03 23:46:22 +03:00
8ab4649fef
FIX Flips sign of SVM's dual_coef_ in binary case
...
Makes it consistent with other sign adjustments, adds
a test to reproduce decision_function in a kernel case
2015-03-03 02:29:23 +03:00
354c35cd5b
FIX Make LinearSVC/LinearSVR support the deprecated l1/L1, l2/L2
2015-02-25 19:12:27 +05:30
50876eca47
COSMIT Flake8 fixes for svm/base.py
2015-02-25 19:12:27 +05:30
9df01a8a4e
FIX/MAINT Deprecate and support upper case values for loss/penalty
2015-02-25 15:34:56 +05:30
c1c8cc1e3a
FIX loss, penalty --> penalty, loss
2015-02-25 15:34:56 +05:30
dbc5d707c6
Change loss names for LinearSVC and LinearSVR()
...
the names are now consistent across methods
In LinearSVC:
'l1' -> 'hinge'
'l2' -> 'squared_hinge'
In LinearSVR:
'l1' -> 'epsilon_insensitive'
'l2' -> 'squared_epsilon_insensitive'
2015-02-09 11:09:16 +01:00
b3fbccca38
FIX various mismatch between docstring and signature params
...
DOC max_iterations -> max_iter. Make it consistent with kmeans
MAINT Replace the deprecated dx parameter with d in the docstrings
MAINT Deprecation warning for max_iterations parameter.
2015-01-16 11:54:57 -05:00
abd31d2cad
MAINT Make uniform the error raised for not fitted condition
2015-01-12 00:33:40 +05:30
8e5bd8ef35
COSMIT / PEP8 Limit line length to 79
2015-01-12 00:32:20 +05:30
7e1a0eee08
ENH LinearSVR using liblinear code
...
Fixes gh-3877.
2014-12-03 20:44:04 +01:00
4d1692c9a2
FIX sign flip in (Nu)SVR with linear kernel
...
Fixes #2933 .
2014-10-12 17:51:40 +02:00
b1119bbb74
ENH friendlier message for calling predict before fit on SVMs
...
Fixes #3601 .
2014-09-03 12:56:57 +02:00
e77b8ef747
MAINT remove harmless, but useless, double call to csr_matrix
...
The input validation code will already have performed the conversion.
Calling this constructor on a CSR matrix doesn't actually copy, but
it's useless too.
2014-09-03 12:00:32 +02:00
a7ccbb8096
MAINT: Made the following changes
...
1. Use a single loop for both ovr and multinomial conditions
2. Better warning messages for penalty and loss conditions
3. Convergence warnings print only if verbose > 0
4. Used check_X_y instead of check_X
2014-08-30 12:26:40 +02:00
34a4517983
ENH: Improved verbosity and denesting
2014-08-26 20:48:23 +02:00
959a4ac698
MAINT: Remove BaseLibLinear for LogisticRegression
2014-08-26 13:58:07 +02:00
4cd177758f
ENH: Merge MultinomialLR into LR with multi_class='multinomial'
2014-08-21 17:37:37 +02:00
ad5773cfca
TST: Added test to check ConvergenceWarning
2014-08-01 12:15:57 +02:00
c8c72fd96e
ENH: Return n_iter_ from liblinear and print convergence warnings
2014-08-01 12:15:57 +02:00
3edfa3cf17
Max iter param in Liblinear is now softcoded
2014-08-01 12:15:57 +02:00
a52174b22c
FIX: Changes due to recent refactoring of the check functions
2014-07-22 16:57:41 +02:00
0634ddf500
FIX: Class weights are computed for each OvA
2014-07-22 16:57:41 +02:00
MechCoder