[scikit-learn] annotated tag debian/0.16.0_b1+git1-gab4d07d-1 created (now e87b6e7)
Andreas Tille
tille at debian.org
Wed Dec 28 13:11:06 UTC 2016
This is an automated email from the git hooks/post-receive script.
tille pushed a change to annotated tag debian/0.16.0_b1+git1-gab4d07d-1
in repository scikit-learn.
at e87b6e7 (tag)
tagging c8d3df4c97fb13d0952d2dd1b23d184198505d19 (commit)
replaces debian/0.15.2-3
tagged by Yaroslav Halchenko
on Wed Mar 11 21:25:31 2015 -0400
- Log -----------------------------------------------------------------
scikit-learn Debian release 0.16.0~b1+git1-gab4d07d-1
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1
iEYEABECAAYFAlUA6wsACgkQjRFFY3XAJMjeXQCgpWubHutwRUw+HeayJwGw2HJb
TB4An0JJgP1DSlFDDtEukQbAKnNtNqDG
=lMQ4
-----END PGP SIGNATURE-----
A. Flaxman (1):
DOC: copy-edits and active voice
Aaron Schumacher (2):
typos: remove extra s's
tweak to more natural version
Aaron Staple (1):
Add quantile strategy to DummyRegressor. Fixes #3421.
Akshay (1):
Removed n_jobs parameter from the fit method and added it to the constructor
Aldrian Obaja (1):
FIX bug in fit_params assumed to be an array
Alexander Fabisch (4):
Remove trailing '\' and add test
Describe default metric
Add t-SNE to whatsnew
Fix issue #4154
Alexandre Gramfort (69):
Merge pull request #3343 from ihaque/doc_partial_fit
Merge pull request #3344 from ihaque/remove_dup_fit
Merge pull request #3368 from stevetjoa/typo-diriclet
Merge pull request #3375 from ldirer/hashing_fix3356
Merge pull request #3472 from arjoly/fix-metrics-division
Merge pull request #2862 from MechCoder/LogCV
Merge pull request #3465 from pvnguyen/patch-1
Merge pull request #3588 from tejesh95/patch-1
fix_gnb_proba
Merge pull request #3653 from MechCoder/minor_opt
rephrase doc, rename unseen, fix pep8
Merge pull request #3674 from MechCoder/fix_l1_penalty
Merge pull request #3697 from luispedro/cv_doc_fix
Merge pull request #3705 from floydsoft/patch-1
Merge pull request #3724 from mlopezantequera/patch-1
Merge pull request #3740 from MechCoder/lassolarsicattribute
cosmit in warning message
Merge pull request #3480 from dsullivan7/sgd
Merge pull request #3770 from dsullivan7/asgdintercept
Merge pull request #3773 from dsullivan7/sgdtestfix
Merge pull request #3646 from s8wu/multinomial_newtoncg
Merge pull request #3778 from MechCoder/expose_positive
Merge pull request #3783 from jakevdp/gmm-doc-fix
Merge pull request #3772 from MechCoder/manhattan_metric
Merge pull request #3795 from dsullivan7/svm_doc
Merge pull request #3811 from MechCoder/fix_repeated_checking
Merge pull request #3861 from ogrisel/fix-python-2.6-test
Merge pull request #3857 from MechCoder/test_sparse_knc
Merge pull request #3867 from snuderl/patch-1
Merge pull request #3868 from MechCoder/typo_example_faces
Merge pull request #3889 from jdcaballero/master
Merge pull request #3905 from jlopezpena/fix-example-plot_pca_3d
Merge pull request #3920 from jlopezpena/pls-coef
Merge pull request #3923 from Lothiraldan/patch-1
add example for birch + make it more pythonic
Merge pull request #3940 from ragv/doc_own_estimators
Merge pull request #3954 from amueller/fix_sgd_learningrate_docs
Merge pull request #3802 from MechCoder/birch
Merge pull request #3962 from MechCoder/check_X_y
Merge pull request #3931 from dsullivan7/cwwarn
Merge pull request #3964 from bwignall/boostdoctypo
Merge pull request #3824 from nmayorov/rfecv_bugfix
Merge pull request #3969 from MechCoder/remove_check_X_y
Merge pull request #3966 from mvdoc/wtree_children
rename alpha to shrinkage + docstring cosmits
use random state in test (no global seed) + cosmit
Merge pull request #3974 from lesteve/fix-silhouette-typo
Merge pull request #3979 from lmichelbacher/patch-1
Merge pull request #3952 from MechCoder/callable_connectivity
Merge pull request #3985 from MechCoder/return_distances
Merge pull request #3997 from pricingassistant/doc-typo-fix
Merge pull request #3994 from jnothman/vec_dbscan
Merge pull request #4010 from MechCoder/sparse_bug
Merge pull request #4012 from pricingassistant/doc-typo-fix
Merge pull request #4021 from jakevdp/fix_check_symmetric
Merge pull request #4024 from jakevdp/ensure_symmetric
Merge pull request #4044 from andreasvc/patch-1
FIX : avoid unecessary computation sparse dual gap
Merge pull request #4105 from amueller/minor_spelling_fixes
Merge pull request #4113 from ragv/isotonic
Merge pull request #4118 from jmetzen/kernel_ridge_doc
Merge pull request #4109 from amueller/gmm_tied_covariance
Merge pull request #4229 from ragv/4135
Merge pull request #4244 from ugurcaliskan/patch-1
Merge pull request #4214 from ogrisel/fix-empty-input-data
Merge pull request #4204 from mvdoc/master
Merge pull request #4265 from harrymvr/doc_spectral_clustering
COSMIT : pep8 + 2 spaces
TST improve coverage of calibration.py
Alexis Mignon (7):
Change the way the covariance is computed to avoid problem with not positively defined covariance matrices
Added test to check that obtained covariance matrices are positive definite after learning a GMM
Cosmetic changes in the doc string. Added an indentation level
Clarified a comment and factorized the tests for all covariance types
Converted test_positive_definite_covars to a generator.
Converted docstring of 'test_positive_definite_covars' into a comment to avoid description ptoblems with nose (Related to issue #4250)
Put the error state change about underflow errors in a 'with' statement in '_covar_mstep_full'.
Anders Aagaard (1):
FIX six issue with module imports
Andreas Mueller (171):
TST add test for minibatch k-means reassigns.
FIX bug where random state was reset in each call to partial_fit, such that reassignment either occured never or always.
FIX bug where distances were of illegal shape
FIX crash when to_reassign.sum() > X.shape[0]
TST split test for partial_fit and fit
TST add additional test for batch_size > n_samples in fit.
MISC rename random.pyx to _random.pyx
ENH backport np.random.choice.
FIX broken tests for minibatch k-means
FIX use choice in minibatch_kmeans
FIX another (two?) bugs: the default code path didn't ever compute distances.
skip doctests
simplify tests
remove redundant code
FIX up the rest of the stuff. Introduce .5 as a magic number. Hurray for magic numbers?
DOC more / better documentation
FIX reset counts of reassigned clusters.
ENH never reassign everything, more meaningful tests.
[DOC] minor fixes
ENH allow y to be a list in GridSearchCV, cross_val_score and train_test_split.
add test for safe_indexing, add another test for cross_val_score
BUG: Support array interface
Catch ConvergenceWarning in RandomizedL1
ENH rename parameters in MockListClassifiers.
move around examples for better structure.
Add related project to website.
ENH don't convert dataframes in grid search and cross-validation
split test_common.py into checks and test file.
Closes #2360. Fix tiebreaking.
some cleanups in common_test, speedup.
Make fit_transform and fit().transform() equivalent in nmf
slight speedup
Make everything accept lists as input.
ENH Make all decision functions have the same shape. Fixes SVC and GradientBoosting. Closes #1490.
Refactor input validation.
remove check_arrays stuff and old input validation
ENH add allowed_sparse named argument for @ogrisel
Merge pull request #3447 from amueller/input_validation_b
DOC Add solido testimonial
Update feature_stacker.py
FIX in KDE fit and score allow y=None, add test for use with pipeline and gridsearch
Fix using ground truth in silhouette score in document clustering. Closes #3806.
DOC explain SelectorMixin strategy with multiple classes.
DOC Add note about sample weights in export_graphviz.
FIX raise error when number of features changes in partial fit, Add common test for inconsisten number of features for partial fit.
DOC RidgeCV more explicit documentation of CV parameter, taken from GridSearchCV.
Remove unused distance calculation from clustering example.
DPC FIX latex error in multiclass hinge loss
Update documentation.rst
WEB fix css for new version of sphinx.
Use plot_galler=0 in makefile for sphinx >1.2.3 Also works with 1.2.3
Merge pull request #3892 from mrshu/mrshu/naive-bayes-small-fixes
Merge pull request #3921 from trevorstephens/docstring-format-fixes
DOC add missing import to plot_underfitting_overfitting.py, fix plot height.
Merge branch 'pr/3914'
FIX random state in isomap test
Allow list of strings for type_filter in all_estimators.
DOC explain t0 in the docstring (currently somewhat confusing)
Merge pull request #3955 from hammer/patch-1
remove _iff_attr_has_method function
don't give names to decorators
FIX backward compatibility.
Merge pull request #3523 from cle1109/slda
make attribute decriptor have the right docstring.
add regression test for docstring delegation
fix docstrings in sphinx
rename decorator
Fix typos, no shortenings in name, simplify example.
TST add tests for nd grid-search can train_test_split
remove deprecated ``n_bootstraps`` parameter in Bootstrap cv object, verbose in FactorAnalysis and SelectorMixin. Also adjust some deprecation strings.
remove deprecation tests.
Add parameters to pipeline methods, add some missing docstrings.
If scoring is provided, it MUST be used. Also, a scorer might not need predict.
tighten tests and adjust for new logic.
add some nonregression tests.
add fixes to whatsnew
remove bad estimator test as it was enforcing too many restrictions, instead test that we don't enforce these restrictions.
make gridsearch_no_predict test stricter.
add another bug that I just discovered existed and was fixed in the PR to whatsnew.
address @jnothman's comments
fix docstring for pipeline score func.
Merge pull request #4049 from akshayah3/sparse
Merge pull request #4054 from tttthomasssss/bugfix-4051-affinity-prop
Say best_estimator_ depends on refit=True. Fixes #2976.
Merge pull request #4078 from pratapvardhan/misc-docs
TST slight cleanup of common tests.
Merge pull request #4075 from banilo/pipe_typo
Merge pull request #4089 from wujiang/patch-1
Merge pull request #3640 from untom/fix_consensus_score
does not compute
make related packages more prominent
Merge pull request #3562 from arjoly/bench-mnist
Make ParameterSampler sample without replacement if all parameters are given as lists.
TYPO in model evaluation docs
TYPO in coverage_area docstring.
COSMIT spelling
DOC minor fixes do docstrings, don't document deprecated parameters.
TST add test for sparse matrix handling in clustering
Merge pull request #4052 from amueller/clustering_sparse_matrix
TST set random state in check_classifiers_train
remove duplicated test, probably caused by rebase issues.
Explain why we are somewhat selective, lower citiation rule of thumb
FIX check for integer 1 before checking for floating 1 (isinstance(1, numbers.real) == True)
Merge pull request #3836 from untom/fixedsplit
add PredefinedSplit to classes.rst and whats_new.rst
Added input validation refactoring from last sprint to whatsnew. Not padding my back, just keep forgetting if that was in 0.15.2 or not.
DOC staged_* returns generators. Fixes #3831.
TST remove tempfiles that we create in svmlight tests and agglomerative clustering tests
Merge pull request #4155 from trevorstephens/nb-fix-3186
Add test for MSE estimate bug.
sort labels in precision_recall_fscore_support
search C not gamma of linear svm in tutorial
Make nystroem approximation robust to singular kernel.
TST Silence some tests.
minor fixes, addressing @agramfort's comments.
website: add skll to related projects
Merge pull request #4201 from amueller/related_projects_skll
Add test that score takes y, fix KMeans, FIX pipeline compatibility of clustering algorithms!
Explain why we multiply mean by two
Better error messages in MeanShift, slightly more robust to bad binning.
Merge pull request #4176 from amueller/mean_shift_no_centers
Merge pull request #4221 from saketkc/fixes
Merge pull request #4232 from lesteve/use-absolute-imports-in-tests
Fix gibbs sampling behavior in RBM with integral random_state.
FIX check (and enforce) that estimators can accept different dtypes.
fixes in GMM, TSNE, MDS, LSHForest, exclude SpectralEmbedding
DOC minor improvement in Ensemble user guide
Merge pull request #24 from ogrisel/fix-nan-1d-regularized-covariance
Merge pull request #4136 from amueller/test_dtypes_all
Merge pull request #3891 from ragv/decision_function_ovo_1523
make check_array convert object to float.
more robust check for dtype object
extensive (excessive?) testing of FDR
raise noise in test again, be ascii
ENH make defensive copies in GradientBoosting*.staged_decision function.
Merge pull request #4165 from amueller/gbrt_staged_defensive_copies
Merge pull request #4146 from amueller/fdr_treshold_bug_2
special case if dist is object in _get_weights
whatsnew entry for KNeighbors zero division in weights.
Merge pull request #4269 from MechCoder/deprecate_n_components
Merge pull request #4092 from ogrisel/iris-rbf-params-heatmap-example
Merge pull request #4293 from kyleabeauchamp/patch-1
Make BaggingClassifier use if_delegate_has_method in decision_function
remove special case for BaggingEstimator after merge of #4137
TST make TheilSenRegressor run faster in common tests, I could swear I did that before...
Fix docstring of mahalanobis distance in empirical covariance. Closes #4168.
Merge pull request #4310 from larsmans/svm-bounds-l2
skip OMPCV on travis by raising SkipTest in set_fast_parameters.
some fixes for sphinx and in examples
don't use unicode in latex. That just complicates things too much. Also, who the hell uses unicode whitespace?
use PNG images for latex compatibility
Merge pull request #4314 from vortex-ape/RBFSampler
catch some warnings, be less verbose in testing.
Merge pull request #4320 from amueller/minor_doc_fixes
Generate example rst files for inclusion in docstrings.
Merge pull request #4323 from amueller/generate_empty_example_rsts
minor fixes in docs.
change default to shuffle=True in SGDClassifier and friends.
Merge pull request #3965 from amueller/sgd_classifier_shuffle
Merge pull request #4325 from amueller/more_minor_doc_fixes
Merge pull request #4234 from amueller/rbm_random_state_gibbs
Merge pull request #4338 from lesteve/fix-typo-in-pyamg-skip-test-message
Implement "secondary" tie strategy in isotonic.
Merge pull request #4345 from bendavies/20newsgroups_example
add scipy2013 tutorial links to presentations on website.
Merge pull request #4339 from amueller/scipy2013_lecture_link
Merge pull request #4342 from sotte/fix_contributing_format
Fix rebase conflict
Fix #4351. Rendering of docs in MinMaxScaler.
Merge pull request #4352 from amueller/issue-4297-infinite-isotonic_bak
0.16.X branching, version 0.16b1
Andreas van Cranenburgh (1):
add 'MultiLabelBinarizer' to __all__
Andrew Tulloch (1):
[Feature Selection] Fix SelectFDR thresholding bug (#2771)
Andrew Walker (1):
Update plot_kernel_approximation.py
Antony Lee (2):
Accelerate AffinityPropagation.
AffinityPropagation: save memory by reusing tmp.
Arnaud Joly (123):
Update what's new
Update what's new
FIX failing tests due to the change of the underlying sample generator
Merge pull request #3366 from MechCoder/link-to-blog
MAINT Re-order argument to put deprecated one at the end
ENH improve forest testing + avoid *args
ENH improve rand_int and rand_uniform
Merge pull request #3406 from jnothman/minor
ENH add label ranking average precision
DOC write narrative doc for label ranking average precision
DOC FIX error + wording
TST invariance testing + handle degenerate case
flake8
FIX use np.bincount
DOC friendlier narrative documentation
ENH simplify label ranking average precision (thanks @jnothman)
ENH be backward compatible for old version of scipy
Typo
pep8
DOC remove confusing mention to mean average precision
Merge pull request #3412 from brentp/logistic_l1_l2_ex
Merge pull request #3411 from lesteve/scheduled-removal-from-0.15
DOC improve documentatino thanks to @vene and remove mention of relevant labels
ENH add sample_weight support to dummy classifier
DOC update what's new
DOC sample_weight attribute
DOC more intuition about corner case
DOC add documentation to backported function
ENH less nested code
Merge pull request #3430 from amueller/test_list_input
Merge pull request #3433 from lesteve/remove-non-integer-deprecation-warnings
ENH better default for test for SelectKBest and random projection
MAINT tree compute feature_importance by default
pep8
Merge pull request #3439 from arjoly/test-commons
FIX encoding issue
Merge pull request #2804 from arjoly/lrap
DOC update what's new
MAINT deprecate fit_ovr, fit_ovo, fit_ecoc, predict_ovr, predict_ovo, predict_ecoc and predict_proba_ovr
MAINT split sklearn/metrics/metrics.py
ENH + DOC set a default scorer in the multiclass module
MAINT flatten metrics module and avoid nested bicluster module
DOC typo + not forgetting single output case
ENH in sklearn.metrics set bicluster as a submodule of cluster
FIX import
DOC improve documentation and distinguish each module
Merge pull request #3401 from vene/scorer_weights
ENH add a friendly warnings before deleting the file
Merge pull request #3445 from arjoly/flatten-metrics
MAINT move log_loss and hinge_loss to the classification metrics
MAINT use assert_warns
MAINT Fix regression random projections work sparse matrices
Merge pull request #3522 from ugurthemaster/patch-1
DOC add link toward Jatin Shash webpage
DOC update what's new
MAINT simplify covertype benchmark
DOC document copy_X parameter from LinearRegression
MAINT add a deprecation version
Merge pull request #3732 from queqichao/fix_typo
FIX ensure that pipeline delegate classes_ to the estimator
pep8
TST bagging of pipeline of classifier
DOC update what's new
Update what's new
COSMIT reshape for all regression strategy + avoid xrange
COSMIT less nested constant check + use comprehension
Revert unwanted modificatoin
Add sample_weight support to Dummy Regressor
Use np.average instead of np.mean
Update what's new: full name of Staple
MAINT remove deprecated oob_score_
MAINT remove deprecated loss in gradient boosting
DOC add missing public function into the references
Update what's new
ENH Bring sparse input support to tree-based methods
FIX+ENH add min_weight_fraction_split support for sparse splitter
Re-organize code dense splitter then sparse splitter
Simplify call to extract_nnz making it a method
ENH while -> for loop
ENH reduce number of parameters
FIX min_weight_fraction_split with random splitter
FIX min_weight_leaf in best sparse splitter
ENH remove spurious code
cosmit
ENH adaboost should accept c and fortran array
COSMIT simplify function call
ENH expand ternary operator
Revert previous version
ENH move utils near its use
ENH add a benchmark script for sparse input data
Extract non zero value extraction constant
Lower number of trees
wip benchmark
Temporarily allows to set algorithm switching through an environment variable
Benchmark: Add more estimators + uncomment text
FIX duplicate type coercision + DOC fix inversion between csc and csr
Remove constant print
COSMIT add Base prefix to DenseSplitter and DenseSplitter
MAINT refactor gradient boosting code
FIX unlikely pickling error of splitters
TST add a toy test for min_weight_samples_leaf
ENH add coverage multilabel ranking metric
Remove copy paste mistake
DOC improve documentation of coverage
TST make separate tests for coverage + remove redundant tests with commons
DOC clarify how ties are broken for coverage_error
ENH add sample_weight support
DOC typo
TST + FIX ensure that it fails if sample_weight has not proper length
DOC more explicit title
Update what's new
DOC add missing load_svmlight_files to api references
ENH Release gil in feature importance
ENH use threading backend for features importance parallelisation
FIX explicit initialization of normalizer
Update what's new
FIX copy paste mistake
MAINT remove deprecated auc_score function
ENH add a benchmark on mnist
ENH Improve script output display
DOC add performance for all available classifiers
FIX Ensure at least 1 feature is sampled when max_features is a float
FIX ensure that negative float max_features will lead to an error
Barmaley.exe (4):
FIX Flips sign of SVM's dual_coef_ in binary case
Updates whats new and fixes the documentation
Removes RBF calculation by hand
Minor fix for a comment
Ben Davies (1):
use a pipeline for pre-processing feature selection, as per best practise
Benedikt Koehler (1):
Typo
Boris Feld (1):
Fix tinyclues logo in doc/about.rst
Borja Ayerdi (3):
Fix RFE n_features minimum value #3812
Fix RFE n_features minimum value #3812 and make it simpler.
If step is explicitly zero or negative, raise a ValueError exception.
Brent Pedersen (1):
use more interesting range for C in logistic l1 l2 example.
Brian Wignall (1):
CLN: Fix typo in comment
Brooke Osborn (9):
adding avg weights and flag to plain_sgd method
asgd is added
adding test that computes the average sgd
checkpoint for classifier
fixing tests
adding documentation fixes and changing avg to average
fixing documentation
adding comments to linear algebra operations
updating doc
CJ Carey (1):
ENH: Use the scipy C-based L1 distance if possible
Calvin Giles (2):
Changed f=open() to with open() as f to eliminate ResourceWarnings. Fixes #3410.
Moved code out of context block where not required for #3612
Cathy Deng (1):
FIX agglomerative clustering w/ connectivity constraints & precomputed distances
Celeo (1):
Fixed typo in CONTRIBUTING.md about how to submit changes back to the project
Christian Stade-Schuldt (1):
TST make catch_warnings blocks more robust
Christof Angermueller (2):
Update documentation of predict_proba in tree module
Update docstring predict_proba()
Chyi-Kwei Yau (1):
fix import error in lda.py
Clemens Brunner (54):
Refactored LDA module (first working version).
Added LabelEncoder and updated most of the docstrings
Updated more docstrings
Return self in fit()
Use equal priors by default
Added simple LDA example
Moved some common code into helper function
Support manually set shrinkage parameter (alpha)
For now only the svd solver supports the transform method
Updated LDA unit tests
Simplified LDA vs. shrinkage LDA example
Removed magic number
Addressed @mblondel's comments
Updated parameter name in help text
Parameters are now consistently named; also refactored the _means_cov() function
Added different shrinkage values to test cases
Updated LDA example and removed decision plane comparison
Increased coverage
PEP8 style conventions
Catch NotImplementedError
Fixed doctest
Use assert_raises() correctly
Removed private _decision_function()
Revert last commit because it broke some tests
Assert that the coefficients of the solvers are approximately the same
Replaced str with six.string_types
Raise error if alpha is not the expected type
Cleaned up docstrings.
Cleaned up one more docstring.
Addressed some comments by @agramfort.
Changed format of shape parameter in docstrings.
Updated test.
Use broadcasting to scale back instead of multiplying with a diagonal matrix.
Added random seed.
By default, estimate priors from sample.
Fixed docstrings.
Updated example because store_covariance is now set in __init__() and not in fit().
Changed parametrization of coef_ and intercept_.
Treat the binary case differently from the case when n_classes > 2.
Introduced step size to speed up the example.
(Hopefully) fixed Travis bug on Python 2.6.
Another try to fix the Travis bug.
Inherit from LinearClassifierMixin.
Added shrinkage LDA entry.
Rename parameter value 'ledoit_wolf' to 'auto'.
Test that lsqr and eigen solvers return almost exactly the same coefficients.
Added deprecation warnings to store_covariance and tol parameters in fit() method.
Added sentence on lsqr vs. eigen solvers.
Now correctly use weighted average to compute the class covariance. Cleaned up code and added references.
Works with one feature.
Updated LDA tests.
Fix problem with np.linalg.norm for version <= 1.6.
Added documentation for the three solvers.
Updated documentation on shrinkage.
Daiki Aminaka (1):
fix typo (on -> one)
Dan Blanchard (1):
ENH sort option for memory-efficient DictVectorizer
Danny Sullivan (51):
more documentation changes
adding test that contains a simple asgd implementation and compares it with the output of sgd_fast
adding more precision to almost equal
converting to setting coef after the fit has been made
putting const on pointer arguments
adding test for binary classifier
adding support for partial fit and adding test for partial fit
adding support for averaged intercept
adding test for intercept part of asgd
adding support for multiclass with a test included
changing float to double
adding comments to xnnz implementation
adding faster implementation of asgd with sparse data
changing api to have average_sgd and plain_sgd
seperating out the average and plain apis in stochastic_gradient.py
adding sgd_fast.c
fixing typos
adding explanation of asgd to doc
adding if self.average logic
adding test to make sure plain sgd does not have average parameters
changing sgd to asgd
adding comparison for ASGD
removing standard_coef and standard_intercept from plain sgd
making optimal for asgd the constant learning rate
fixing merge conflicts
changing averaging to use sparse trick
adding value error for partial fit with auto weights
adding note to whats_new.rst
removing miss-merge and adding github
adding a more descriptive error message
adding escape characters to regex
cleaning up commented out code and previous_coef_ parameter
cleaning up floating point and unneeded todos, also removing constant learning rate for asgd
removing documentation about constant learning rate for averaging
adding add_average method and solving iteration bug by replacing definition of t
adding note for partial_fit and n_iter
clean up and average can now be set to an int indicating the number of iterations before averaging
increasing testing precision, putting comment on one line and adding asgd to regression benchmark
changing parameters for asgd regression benchmark and changing array shape for docstring
adding space to shape and putting // in average division
removing spaces for tuples of width 1
adding asgd to whats_new.rst
fixing merge conflicts
fix to a minor bug with intercept
fixing duplicated classifier in asgd test
adding example for memory wrapper
adding support for class_weight in fit method
adding warning if passing class_weight through fit
changing warn message, making it Deprecation Warning, and removing negative index for py2.7
adding simplefilter for warning
changing warning test to use assert_raises_message
David Fletcher (1):
DOC fix/expand Imputer example docs
Dmitrijs Milajevs (1):
Don’t embed hyperlinks during latex file generation.
Dougal J. Sutherland (2):
fix factor of 2 in RBFSampler; make test more rigorous
clarify KernelDensity.score{,_samples} docstrings
Erich Schubert (1):
Do not shuffle in DBSCAN (warn if `random_state` is used).
Fabian Pedregosa (29):
Cosmetic: correct latex formula display
DOC: write cost function of logistic regression.
DOC: add transpose and intercept to formula of logistic regression
Merge pull request #3378 from kastnerkyle/dense_cholesky_warnings
FIX: check with tolerance on lars_path
Cosmetic np.abs -> abs
DOC: expand documentation for logistic regression
Implementation of logistic_regression_path.
Take into account @agramfort's comments.
Some fixes for LogisticCV object
Docstring of LogisticRegressionCV
Docstring
Refactor and some bug fixing
Docstring
FIX missing import
FIX: bug in LogisticRegressionCV
Make tests deterministic
FIX: coef.shape
Just to be sure
Add test
BUG when fit_intercept=False
Fallback for failing line search
Remove warning (not needed any more)
Compatibility for old scipy
iterate some more on line search
cosmetic
ENH LinearSVR using liblinear code
DOC: make LinearSVR appear in the doc reference
Change loss names for LinearSVC and LinearSVR()
Florian Wilhelm (100):
Added multiple linear Theil-Sen regression
Added an example and documentatin for Theil-Sen
Added subpopulation parameter to Theil-Sen
Added parallelization support to TheilSen Estimator
Improved parallelization for Theilsen estimator
Merge branch 'master' into theilsen
Cleanups and corrections for Theil-Sen regression.
Removed subpopulation=None option in TheilSen
xrange fix for Python3 in TheilSen
FIX Theil-Sen unittest for older Scipy Version
FIX that some functions in Theil-Sen were public
FIX usage of linalg from Scipy in Theil-Sen
FIX: Let Theil-Sen handle n_samples < n_features case
FIX: Python 2.6 format syntax in Theil-Sen
Vectorization of theilsen._modweiszfeld_step
FIX: Parallel unittests for Theil-Sen estimator
FIX: TheilSen supports old Numpy versions
DOC: Comparison of Theil-Sen and RANSAC
DOC: Fixed typo in Theil-Sen example.
FIX: Some coding style fixes in TheilSen unittest.
FIX: Reduced the runtime of the TheilSen unittest.
DOC: Small corrections in the docs of Theil-Sen
DOC: Explanation when TheilSen outperforms RANSAC.
Merge branch 'master' into theilsen
Merge branch 'master' into theilsen
Fix for old Numpy 1.6.3
Added to comments to better explain last commit
Use string argument for legend's loc parameter
Merge remote-tracking branch 'gvaroquaux/pr_2949' into theilsen
Merge remote-tracking branch 'upstream/master' into theilsen
TST: Cleanups in test_theil_sen
COSMIT: Renamed _lse to _lstsq in theil_sen.py
ENH: Removed shared-memory parallelism in theil_sen
COSMIT: Inlined two methods in theil_sen.py
ENH: Use warnings instead of logging in theil_sen
ENH: Removed _split_indices method in TheilSen
ENH: Rewrote TheilSen._get_n_jobs as a function
COSMIT: More explicit names for vars in theil_sen
Merge branch 'master' into theilsen
FIX: usage of check_array in theil_sen
FIX: Use check_consistent_length in theil_sen
ENH: Refactoring in theil_sen
ENH: Removed unnecessary generator in theil_sen
FIX: doctest of get_n_jobs
ENH: Theil-Sen vs. RANSAC example
Merge branch 'master' into theilsen
COSMIT: Small changes regarding Theil-Sen
DOC: Better documentation for Theil-Sen
ENH: Improvements in the Theil-Sen regressor
ENH: Shortcut for 1d case in spatial median
ENH: Avoid trailing \ in test_theilsen imports
FIX: TheilSen -> TheilSenRegressor in docs
DOC: Narrative doc for median_absolute_error
ENH: Reworked _modified_weiszfeld_step
DOC: Improved _spatial_median docs
COSMIT: Replaced xrange by range
COSMIT: Renamed y -> x_old in _modified_weiszfeld_step
COSMIT: Renamed spmed[_old] to spatial_median[_old]
COSMIT: Break and for .. else in _spatial_median
ENH: Reworked _lstsq in theil_sen.py
ENH: Replace AssertionError by ValueError
COSMIT: Improved error message in theil_sen.py
COSMIT: Renamed n_all to all_combinations in theil_sen.py
COSMIT: Consistent naming for n_subpop
COSMIT: Fixed pep8 problem in theil_sen.py
DOC: Moved notes section to long description
Merge branch 'master' into theilsen
ENH: Added median_absolute_error to metrics
DOC: Added doc in model_evaluation.rst
ENH: unit tests for median_absolute_error
DOC: Added return doc to _lstsq
COSMIT: Better variable names for _modified_weiszfeld_step
COSMIT: Moved epsilon to module level
COSMIT: Some empty lines for better readability
ENH: test_common.py unit tests pass
FIX: Small typo
Merge branch 'master' into median_absolute_error
ENH: Removed median_absolute_error due to PR #3761
ENH: Made n_subpopulation a fit parameter
COSMIT: Some renamings and PEP8 compliance
Merge branch 'master' into theilsen
ENH: Removed 1d shortcut in _spatial_median again
COSMIT: Clearer slicing syntax in _modified_weiszfeld_step
ENH: Fixed confusing X = X.T renaming
COSMIT: Some renamings in _lstsq
FIX: Fix of last merge with master
COSMIT: Renamings for easier understanding
COSMIT: Another slicing syntax cleanup
Revert "ENH: Removed 1d shortcut in _spatial_median again"
ENH: sample without replacement
Merge branch 'master' into theilsen
COSMIT: pep8 and renaming
COSMIT: replaced assert by assert_less/greater etc.
TEST: No console output during unit tests
ENH: Always set random_state in unit tests
ENH: Speedup of unit tests
COSMIT: Better consistency
ENH: Added random_state in plot_theilsen.py
MAINT remove multi-output support from meadian absolute error
DOC: Added whats_new for TheilSenRegressor
Félix-Antoine Fortin (1):
DOC: Remove extra # from url in fit.
Gael Varoquaux (93):
DOC: improve Thiel-Sen vs RANSAC example
ENH: speed up theilsen
Merge pull request #3333 from ogrisel/maint-skip-download-tests
Merge pull request #3357 from houbysoft/patch-1
Merge pull request #3369 from NelleV/install_doc
Merge pull request #3353 from kastnerkyle/skip_ompcv_travis
MAINT: more explicit glob pattern in doc generation
DOC: more readable whats_new
DOC: fix classes not referenced in docs
TEST: metrics sample_weight test: almost_equal
Merge pull request #3381 from kastnerkyle/skip_omp_cv_regressors
TEST: fix test failing due to numeric instability
COSMIT: address misc comments
Minibach k-means: change reassignment to uniform
MISC: avoid deprecation warning
MISC: minor changes in MBKmeans
DOC: spelling and phrasing
MISC: better formulation
Merge pull request #3363 from ogrisel/appveyor-ci
Merge pull request #3376 from GaelVaroquaux/fix_mb_kmeans
Merge pull request #3382 from ogrisel/install-doc-update
BUG: fix windows pointer size problem
FIX: try to get windows working
Merge pull request #3385 from GaelVaroquaux/fix_intp
MAINT: minor mailmap update
MISC: avoid overridding figure in benchs
MISC: <= rather than < in tol check (KMeans)
Merge pull request #2694 from amueller/allow_y_lists
BUG: Support array interface
ENH: enable y to only implement the array interface
COSMIT: fix indentation
BUG: fix validation bug
ENH: support non-ndarray subclasses in supervised estimator
ENH: transformers work on non ndarray subclasses
DOC: more comments
COSMIT:
FIX broken test
Merge pull request #3414 from lesteve/scheduled-removal-from-0.15
Merge pull request #3418 from arjoly/dummy-weight
ENH: add a median absolute deviation metric
DOC: add an example of robust fitting
TST: fix doctest
DOC: better documentation for robust models
API: naming: CamelCase class -> camel_case function
Merge pull request #3444 from arjoly/explain-default-score
Merge pull request #3442 from arjoly/split-metrics-module
TST: fix tests on numpy 1.9.b2
Merge pull request #3443 from amueller/input_validation_refactoring
DOC: Fix class link
TST: Fix warnings in np 1.9
DOC:Typo in doc
Merge pull request #3438 from hamsal/sprs-out-dmy
MAINT: add license info
MAINT: update joblib to latest release 0.8.3
MAINT: better gitignore
MAINT: be robust to numpy's DeprecationWarning
Merge pull request #3598 from GaelVaroquaux/np_warnings
Merge pull request #3604 from rahiel/patch-1
DOC: more explicit title
BUG: n_samples instead of n_features in cd_fast
ENH: minor speed-up in k-means
ENH: Remove unused copy
COSMIT: explicit comment
Merge pull request #3608 from agramfort/fix_gnb_proba
Merge pull request #3580 from jnothman/gridsearch-score
Merge pull request #3611 from ogrisel/fix-isfinite-windows
Merge pull request #3564 from jnothman/bounds
Merge pull request #3587 from jnothman/transforms-doc
Merge pull request #3678 from akshayah3/LinearReg
Merge pull request #3730 from ogrisel/fix-libsvm-random-seed
Merge pull request #3763 from ogrisel/fix-warning-test-affinities
Merge pull request #3797 from Titan-C/clean
Merge pull request #3854 from justmarkham/doc-changes
Merge pull request #3862 from Titan-C/doc_comment
Merge pull request #3852 from lesteve/add-cds-funding
COSMIT: minor cleanup
Merge pull request #3690 from queqichao/fix_bug_in_cross_validation_when_using_sparse_matrix_for_fit_params
Merge pull request #3823 from ragv/omp-3644
Merge pull request #3900 from dimazest/latexpdf
Merge pull request #3869 from borjaayerdi/rfe_n_features_to_remove
Merge pull request #3936 from sethdandridge/typofix
Merge pull request #3953 from Titan-C/savefigarg
Merge pull request #3158 from mvdoc/clustering
Merge pull request #3986 from amueller/remove_allow_nd_in_train_test_split
Merge pull request #4042 from lmichelbacher/patch-2
Merge pull request #3961 from trevorstephens/rf-class_weight
Merge pull request #4097 from amueller/remove_unneccessary_initialization
Merge pull request #4111 from ragv/slinear_isotonic
Merge pull request #4159 from aflaxman/gp_docs
Merge pull request #4174 from ragv/fixes_4173
Fix rst formatting
DOC: fix path
Merge pull request #4203 from gwulfs/rel_projs
Garrett-R (2):
Edited _get_weights to make it robust to zero distances
Avoid divison by zero warning in KNeighbors estimators
Gilles Louppe (10):
Merge pull request #3334 from ami-GS/external
MAINT: Effectively remove deprecated parameters
Merge pull request #3715 from trevorstephens/correct-gbc-docstring
Merge pull request #3731 from ndawe/master
Merge pull request #4026 from akshayah3/master
FIX: ensure that feature importances are properly scaled
Merge pull request #4139 from MSusik/negative_njobs
Merge pull request #4190 from trevorstephens/refactor_cw
FIX: add except* in Splitter.init definitions
Merge pull request #4282 from glouppe/tree-fix-4281
Hampus Bengtsson (1):
5 of 6 are zeros, not ones.
Hamzeh Alsalhi (61):
Modified sparse OvR to handle sparse target data
FIX Support unseen labels LabelBinarizer and test
Test sparse target data with dummy classifier
Fit dummy classifier with sparse target data
Op directly on y.data in fit, xrange -> range, y.tocsc
Share array intialization sparse and dense fit
Reorder fit and share denes sprs, Scafold sprs pred
Remove temporary code from test_sparse_target_data
remove zero from random uniform return data
Emulate sample weight support from dense (Rebase)
Add four tests for sparse target dummy, one for each strategy
Correct the class priors computation in the sparse target case
Remove redundant indices, data, indptr appends in predict
Test sparse stratifed corner case with 0s
Combine dense column check in fit
Update y with y.tocsc (Not an inplace op)
Select nonzero elements in predict stratified case
Implement uniform and stratified w/ sparse random choices, optimize concats
np.random.choice -> choice (from utils/random)
Use array for sparse matrix construction, fix constant check
Replace np.random.choice with utils.random.sample_without_replacement
Raise warning unifrom fit sparse target, Raise error uniform precit sparse target
Support matrix output random_choice_csc, Update usage in dummy predict
Validate lengths equal each element of classes and class_probabilities
Test random_choice_csc
Test ValueError for length mismatch between array in classes, and probs
Absract class prior construction to function sparse_class_distribution
Test sparse_class_distribution, Correct data indexing
Move array intiialiaztions into dense conditional, nz_indices -> col_nonzero
Rever formating of warning, Reword warning for sparse uniform case
Correct spell 'Predicting', doument (sparse_target_input_ -> sparse_output)
Reverse conditionals in fit and predict, positive case first
Make cosmetic revisions to random_choice_csc
import divison from future to give real results with two ints
Make cosmetic changes to sparse_class_distribution
Make cosmetic changes to sparse_class_distribution
Make naming changes in sparse_class_distribution
Make default parameters for test_random_choice_csc, and use almost equal
move random_choice_csc to utils/random from utils/sparsefuncs
Move test_sparse_class_distribution to utils/multiclass from utils/sparsefuncs
Uodate imports to correct function loactions
Comment .eliminate_zeros() in sparse class distribution
Include dense case support in class_distribution, pep8 revisions
Make sparse target dummy tests multioutput-multiclass
Test class_distribution w/ multioutput-multiclass sparse and dense
Replace numpy.random.choice with a search sorted strategy for faster runtime
Test random_choice_csc implicit, readability adjustments
Remove transposes from testing of random_choice_csc
Clarify unfiorm sparse warning message, Reword sparse_output_ doc
Clarify unfiorm sparse warning message, Reword sparse_output_ doc
Use UserWarning in place of SparseEfficiencyWarning
Remove outer for loop in dense case predict
Make cosmetic changes
Test that class dtypes string and float fail in random_choice_csc
Test explicit sample wieghts in the sparse case of class_distribution
Test insertion of 0 class in random_choice_csc
Change dtype check conditional to look for everything other than int
Make cosmetic adjustments, Fix random.py header, use message with assert_warns
Test additional corner cases with random_choice_csc, error on proabilites not summing to 1
Combine sparse and dense test for class_distribution
Manage explicit zeros manualy in class_distribution, test with explicit zeros
Hanna Wallach (2):
FIX issue #4268 (bug in BernoulliNB).
TST check that Bernoulli NB is consistent with Manning et al. IR book example
Harry Mavroforakis (1):
DOC: specifies the default number of eigenvectors
Hasil Sharma (1):
FIX Make Spectral Embedding deterministic
Helder (1):
Fix a typo on svm.rst
Herve Bredin (8):
ENH: improve GMM convergence check #4178
ENH: issue #4178 (cont.)
FIX: fix doctest
doc: mention GMM stopping criterion change
a few changes according to comments by @ogrisel
ENH: better convergence check
DOC: what's new about GMMs
DOC: make it clear 'thresh' should be removed in v0.18
Hsiang-Fu Yu (1):
MAINT some cleanup in Liblinear
Hugues SALAMIN (1):
Fix for overflow warning
Ian Gilmore (1):
ENH added digits as optional arg for classification_report
Ilambharathi Kanniah (2):
DOC adding return type and improving module doc
DOC return types in datasets.lfw
Imran Haque (5):
ENH Add partial_fit to GaussianNB
Update Sphinx docs for GaussianNB partial_fit
Remove duplicate GaussianNB.fit() code
Preserve public API
docstring
Jake VanderPlas (8):
DOC: fix doc inconsistency in GMM
BUG: only symmetrize matrix when it is not already symmetric
MAINT: create ensure_symmetric utility function to check matrix symmetry
rename ensure_symmetric -> test_symmetric
TST: add test of check_symmetric
DOC: fix docstring of check_symmetric
add more informative doc & error message in utils.check_symmetric
make utils.check_symmetric future-proof
Jan Dlabal (2):
Fix grammar
Fix labels of inliers vs outliers
Jan Hendrik Metzen (37):
FIX Using random_state.rand rather than scipy.rand consistently in Gaussian Process
TST regression test for optimum over several random-starts of GP
TEST A duplicate minimum value should not yield non-finite predictions in IsotonicRegression
FIX Adding eps to minimum value in clipping in IsotonicRegression
DOC Add inline comment with reference to scipy issue
ADD @mblondel's initial implementation of KernelRidge added
ADD Example comparing kernel ridge and support vector regression
TEST Added tests for KernelRidge based on @mblondel's code in lightning
DOC Added documentation for kernel_ridge module
ENH Example comparing KRR and SVR documented and slightly modified
FIX Fixed assigning colors to methods in KRR example
DOC Added narrative doc for kernelized ridge regreesion
DOC Resolved minor issues in documentation of kernel_ridge
MISC Optimized two numpy statements in kernel_ridge.py
FIX Add kernel_ridge to list of all submodules
FIX Input validation in fit() of KernelRidge
FIX Allow multi-output in KernelRidge
REFACTOR Backported changes in _solve_cholesky_kernel to ridge.py
MISC Using RandomState object instead of setting global seed
DOC Revised documentation of kernel ridge regression
DOC Polishing doc of kernel_ridge
PEP8 Removing PEP8 violations in plot_kernel_ridge_regression.py
MISC Let sample_weight default to None
FIX We must not use overwrite_a=True in _solve_cholesky_kernel as K might be reused
TEST Extending tests for kernel_ridge
REFACTOR Using lstsq rather than manual pinv in _solve_cholesky_kernel
MINOR Fixed typo in test name (test_kernel_ridge_singular_kernel)
TEST Use make_regression instead of make_classification when testing kernel_ridge
DOC Added KernelRidge to whats_new.rst
FIX KernelRidge checks if fitted before predict
DOC Added kernel_ridge to classes.rst
ENH Add probability calibration based on isotonic regr. and Platt's sigmoid fit + calibration-curve
ENH Brier-score loss metric for classifiers
TST Tests for the calibration module
DOC Examples for the calibration of predicted probabilities and calibration-curves
DOC Narrative doc for the calibration module
TST Adding brier_score_loss to test_common.py
Jatin Shah (2):
Add sample_weight parameter to metrics.jaccard_similarity_score
Add sample_weight parameter to metrics.log_loss
Javier López Peña (3):
Fix example in plot_pca_3d.py (color array had wrong size)
Check if targets is a numpy array and convert it into one if it isn't
Replace 'coefs' by 'coef_' in PLSRegression
Jean Kossaifi (2):
Minor doc enhancement: documented setup develop.
Added link to the setuptools doc and a note on the need to rebuild every time a
Jeff Hammerbacher (2):
'none' is an acceptable value for penalty
Small typo fix in the comments: "whther" --> "whether"
Joel Nothman (128):
ENH make_multilabel_classification for large n_features: faster and sparse output support
COSMIT in response to @arjoly's comments
COSMIT pep8 fixes
TST ignore sequence of sequences DeprecationWarnings
DOC link example gallery scripts rather than inline
COSMIT
DOC show referring examples on API reference pages
DOC ensure longer underline
DOC mention doc-building dependency on Pillow
Merge pull request #3327 from jnothman/examples_in_apiref
DOC make neural networks example appear
DOC Fix example path
DOC fix 'Return' -> 'Returns'
FIX Py3k support for out-of-core example
DOC add links to github sourcecode in API reference
DOC fix opaque background glitch when hovering example icons
DOC fix doc errors in utils.testing
DOC fix formatting of attributes etc. in docstrings
DOC fix styling of See Also sections
DOC fix styling of method signatures
DOC fix see also references
DOC move Attributes section to after Parameters and style likewise
Merge pull request #3491 from ogrisel/fix-warning-test-cross-val
DOC fix see also reference
DOC fix markup error
ENH Sparse multilabel target support
DOC label indicators are clearer as ints than floats
DOC more precise input type for multilabel metrics
Merge pull request #3509 from arjoly/fix-rp-sparse
Merge pull request #3489 from jnothman/attributes-doc
DOC remove backticks from around attribute name + PEP8
DOC A less-nested coverage of model evaluation
Merge pull request #3528 from MechCoder/fix_heisen
Merge pull request #3527 from jnothman/unnest-model-evaluation
FIX only use testing.ignore_warnings in tests
ENH faster safe_indexing for common case
Merge pull request #3539 from jnothman/fast_indexing
DOC fix typo
DOC fix typo
DOC specify X shape for precomputed
DOC correct default value
DOC add missing details to what's new
DOC correct shape of Tree.value
COSMIT update sklearn.svm.bounds
MAINT Remove note that assert_warns comes with Numpy 1.7
Merge pull request #3572 from AndrewWalker/patch-1
ENH more explicit error message for ill-posed problem
DOC more explicit parameter descriptions in make_multilabel_classification
FIX PLSRegression again supports 1d target
FIX out-of-core example had been broken
DOC avoid the plot_ prefix in example where no plot
DOC fix line references in tutorial
FIX more intuitive behavior for *SearchCV.score
DOC document MultiLabelBinarizer.sparse_output param
Merge pull request #3586 from MechCoder/python3_installation
DOC fix link embedding regexp
DOC allow link embedding to proceed when web sites unavailable
DOC extend documentation on sample generators
Merge pull request #3001 from jnothman/doc_sample_gens
Merge pull request #3593 from MechCoder/row_norms_doc_fix
Merge pull request #3591 from MechCoder/refactor_logreg
DOC tweak BaseSearchCV.score docstring notes
DOC a further tweak to BaseSearchCV.score docstring
FIX update sklearn.__all__ to include all end-user submodules
DOC restructure data transformation user guide
DOC fix broken URL
Merge pull request #3676 from danfrankj/master
DOC reduce table width through less-qualified function names
Merge pull request #3742 from abhishekkrthakur/master
FIX score precision in doctest
Merge pull request #3571 from jnothman/fix-link-embedding
Merge pull request #3833 from amueller/sgd_partial_fit_variable_length
DOC fix what's new for sparse trees
DOC add documentation improvements to what's new
DOC narrative docs for grid search's robustness to failure
DOC fix layout: ensure stable width of documentwrapper
Merge pull request #3881 from amueller/code_tag_new_sphinx
Merge pull request #3893 from jnothman/css_fix
Merge pull request #3578 from jnothman/pls-validation
DOC more expansive description of well-established algorithms
FIX P/R/F metrics and scorers are now for binary problems only by default
Merge pull request #2679 from jnothman/prf_average_explicit
DOC editing for LSH forest narrative documentation
TST/FIX ensure correct ducktyping for metaestimators
TST extend ducktype testing to handle #2853 case
FIX ducktyping for meta-estimators
TST more rigorous testing of delegation
Merge pull request #3989 from hammer/patch-2
ENH support sparse matrices in LSHForest
ENH vectorise _find_longest_prefix_match
TST LSHForest benchmark script uses vectorised queries and fixed random_state
FIX not all scipy.sparse support rand with random_state
STY cleaning some LSHForest query code
COSMIT flat > nested / avoid duplicate validation
STY remove unnecessary variable/operation
ENH sparse metric support for paired distances
Merge pull request #3995 from jmetzen/fix_isotonic
COSMIT Minor changes to address review comments
ENH vectorize DBSCAN implementation
ENH sparse support in DBSCAN
ENH support weighted samples in DBSCAN
FIX DBSCAN.fit_predict now also supports sample_weight
DOC/COSMIT clean up according to review
DOC mention sparse matrix in LSHForest param docstrings
FIX correct name due to rebase error
Merge pull request #4003 from ClashTheBunny/patch-1
DOC add note on scipy.sparse.rand use
Merge pull request #4038 from jakevdp/ensure_symmetric2
Merge pull request #4071 from amueller/grid_search_document_best_estimator
Merge pull request #4029 from ragv/maint_raise_uniform_error_notfitted
Merge pull request #4086 from ragv/whatsnew
Merge pull request #4093 from MechCoder/regresssion_nc
ENH use parallelism for all metrics in pairwise_{kernels,distances}
Merge pull request #4085 from jnothman/pairwise_parallel
Merge pull request #3850 from amueller/randomized_search_no_replacement
DOC what's new on parallelism of pairwise_distances
Merge pull request #4076 from ragv/silhouette_plot
Merge pull request #4158 from ragv/whatsnew
Merge pull request #4160 from trevorstephens/doc-fix-4098
Merge pull request #4131 from amueller/faq_no_1000_citations
TST/FIX in future, average='binary' iff 2 labels in y one of which is pos_label
Merge pull request #4037 from ragv/make_storing_stop_words_optional
Merge pull request #4206 from ogrisel/fix-strict-select-fdr
TST Test check_consistent_length and TypeError with ensemble arg
Merge pull request #4308 from terrycojones/fix-missing-space
Merge pull request #4315 from ugurcaliskan/patch-2
TST ensure return type of radius_neighbors is object
FIX/TST boundary cases in dbscan (closes #4073)
Jonathan Helmus (1):
BUG: Correct check for local import
Kaicheng Zhang (1):
G is upper-triangular, so G.T should be lower-triangular.
Kevin Markham (1):
DOC FIX: typo and minor update
Kyle Beauchamp (2):
Added feature to fix #1523
Update related_projects.rst
Kyle Kastner (10):
Added utility to skip tests if running on Travis
Added FIXME annotation to skipped test
Changed solver from 'dense_cholesky' to 'cholesky' to eliminate deprecation warning.
Additional skipping for omp_cv
Put check_travis in the correct place so that other tests are not skipped, and changed the CCA check to be more explicit.
Added additional skip checking for train and pickle tests
Added directory checking for documentation builds, and corrected for Windows pathing
Merge pull request #3477 from kastnerkyle/docbuild_fix
IncrementalPCA implementation
Updated what's new to add IncrementalPCA
Lagacherie Matthieu (1):
test case on rtfe grid_scores fix
Lars Buitinck (120):
DOC what's new: GaussianNB.partial_fit + typo in comment
DOC double backticks for fixed-width (code) font
TST crank up preprocessing tests to 99% coverage
FIX Gaussian KDE should return array, not scalar
COSMIT use explicit RandomState in KDE tests
DOC more explicit docstring for preprocessing.normalize
DOC optimize some PNG images, shaves off 24kB
DOC run optipng before uploading website
DOC: typo, envelop → envelope
Merge branch 'pr/3358'
DOC numpydoc convention for Bunch-returning functions
Merge branch 'pr/3316'
MAINT ignore sparsefuncs_fast in diffs
MAINT mailmap update
FIX expit bug with out != None
FIX potential overflow in _tree.safe_realloc
FIX DBSCAN input validation
MAINT remove deprecated PCA code
MAINT remove deprecated code from trees and forests
MAINT remove deprecated sklearn.pls module
MAINT remove deprecated code from preprocessing
MAINT remove sklearn.test
COSMIT clean up tests with pyflakes
MAINT remove deprecated code
DOC typo in warning
Merge pull request #3479 from MechCoder/improve_logcv_docs
DOC fix references to moved examples
DOC mistake in comment
Merge branch 'pr/3535'
MAINT NumPy 1.10-safe version comparisons in joblib
Merge branch 'pr/3545'
DOC GP fixes in what's new
Merge branch 'pr/3395'
DOC typo
DOC vectorizers were referring to a private function in public docs
Merge pull request #3573 from MechCoder/high_dimensional_enetcv
ENH multinomial logistic regression using L-BFGS
DOC deprecation of **kwargs in neighbors per 0.18
TST fix doctest for k-NN
Merge pull request #3603 from MechCoder/fix_imputation_example
DOC rename "imputation.py" to "missing_values.py"
DOC dtype parameter on validation functions
DOC multi_output parameter on validation function
MAINT optimize machinalis.png, -9kB/42%
MAINT remove harmless, but useless, double call to csr_matrix
MAINT use %r for better printing of regexps
ENH friendlier message for calling predict before fit on SVMs
MAINT remove unused MST code
FIX division by zero warning in LassoLarsIC
MAINT big refactor of sklearn.cluster.bicluster
Merge branch 'pr/3613'
ENH micro-optimize gradient boosting
MAINT move comment to the appropriate place
FIX don't compare strings with 'is'
FIX propagate MemoryError from Tree._resize
MAINT refactor DictVectorizer's transform+fit_transform (1)
Merge branch 'pr/3683'
TST more robust test for MemoryError from Tree._resize
MAINT: handle frombuffer with empty 1st arg in utils.fixes
DOC single-pass DictVectorizer in what's new
MAINT: set attributes as last action in DictVectorizer.fit
Merge pull request #3687 from perimosocordiae/fast-l1-dist
TST stronger test_shuffle_on_ndim_equals_three
Merge branch 'pr/3696'
Merge pull request #3695 from dougalsutherland/kde-docs
Merge pull request #3656 from dougalsutherland/fix-rbf-samp
FIX ovr predict_proba in the binary case
Merge branch 'pr/3710'
DOC for the Git novices: cd to the clone directory
Merge pull request #3746 from MechCoder/nearest_neighbor_manhattan
DOC fix (Nu)SVR attribute shapes in docstring
FIX sign flip in (Nu)SVR with linear kernel
Merge pull request #3761 from FlorianWilhelm/median_absolute_error
DOC: FAQ: what to do with strings
DOC fix Isomap docstring formatting
Merge pull request #3786 from AlexanderFabisch/tsne_fix
Merge pull request #3800 from MechCoder/unknown_transform
Merge pull request #3829 from amueller/feature_selection_from_model_doc
DOC link to NLTK website
TST stronger non-regression test for #3815
Merge branch 'pr/3837'
Merge pull request #3883 from ogrisel/sgd-stability
Merge pull request #3898 from arjoly/nogil-feature-importance
Merge pull request #3899 from jlopezpena/fix-OvOlist
DOC "RF" -> "Random forest" in text classif example
Merge pull request #3925 from Titan-C/cssbox
DOC typo
MAINT get rid of undefined variable warning in Liblinear
Merge pull request #3934 from amueller/common_test_slight_cleanup
Merge branch 'update-liblinear'
DOC clarify RandomTreesEmbedding docstring
Merge pull request #3968 from MechCoder/metric_neighbors_Graph
MAINT remove probability kwarg from SVR and NuSVR
Merge branch 'pr/3944'
TST remove SVR(probability=True) from AdaBoost tests
COSMIT pep8 fixes to sklearn/tests
FIX LSH benchmark: joblib import, hardcoded /tmp
Merge pull request #3980 from larsmans/lsh-forest
DOC: what's new: LSHForest
Merge pull request #4011 from MechCoder/fix_deprecation_paths
Merge pull request #3982 from amueller/metaestimator_delegation
DOC DBSCAN doesn't "initialize centers"
DOC Birch: spaces
ENH: optimize DBSCAN (~10% faster on my data)
Merge pull request #4151 from larsmans/dbscan-faster
Merge pull request #4153 from hsalamin/remove_exp_warning
Merge pull request #4148 from amueller/clean_up_after_tests
Merge pull request #4141 from amueller/bootstrap_int_one_fix
Merge pull request #4147 from amueller/precision_recall_unsorted_indices
Merge pull request #4156 from AlexanderFabisch/fix_tsne_one_comp
Merge pull request #4171 from amueller/statistical_learning_gridsearch
Merge pull request #4188 from miniyou/patch-1
Merge commit 'b35b2fdad3a38e29bf5ef3534aeb9468b7f20f4a' from #4181
Merge pull request #4198 from arjoly/gbrt-fix-percent-max_features
Merge pull request #4193 from lesteve/deactivate-travis-default-venv
Merge pull request #4179 from amueller/nystroem_singular_kernel
Merge pull request #3933 from fabianp/loss_liblinear
FIX TypeError not ValueError in is_fitted
TST test last commit
MAINT change "l2" to "squared_hinge" in svm.bounds
Laurent Direr (11):
Corrected two typos in docstring.
#3356 - Added an exception raising when np.nan is passed into a HashingVectorizer.
#3356 - Added an exception raising when np.nan is passed into a HashingVectorizer.
Added a test on hashing vectorizer behavior with np.nan input.
PEP8 line length fix.
Replaced assert_raises with assert_raise_message as the point is to make sure the exception message is clear.
Added a comment to explain the use of a test.
Added a parameter check on n_estimators in BaseEnsemble to raise error if it is not strictly positive.
Changed _partition_estimators signature to make it compatible with warm start.
Added warm start to random forests.
Modified the confusion matrix example: included a normalized matrix, changed the colors and added class labels.
Loic Esteve (24):
Remove deprecated 'mode' parameter from sklearn.manifold.spectral_embedding
Remove deprecated 'class_weight' parameter from RidgeClassifierCV.fit
Remove deprecated 'precompute_gram' parameter
Remove deprecated 'copy_Gram', 'copy_Xy' and 'copy_X' parameters
Remove deprecated 'Gram' and 'Xy' parameters from OrthogonalMatchingPursuit.fit
Removed 'Gram' and 'Xy' parameters were still in test_omp.py
Remove deprecated 'score_func' and 'loss_func' parameters from sklearn.metrics.scorer.check_scoring. Amend the code in all the other places they were used.
Remove 'copy_X', 'copy_Gram' and 'copy_Xy' documentation since these parameters have been removed
Tidy up by removing unnecessary local variables
Remove 'DeprecationWarning: using a non-integer number instead of an integer will result in an error in the future' warnings
Remove a couple more 'using a non-integer number instead of an integer DeprecationWarning'
DOC: add funding from Paris-Saclay Centre for Data Science
Added missing GSOC student
MAINT: fix silhouette typo
MAINT create a new venv in travis for "ubuntu"
TST fix tests with numpy 1.6.1
DOC fixes supported numpy version
MAINT improve misleading comment
Move import statement
Simplify the code by using len
MAINT use absolute imports in tests
DOC python 3 fix in plot_kmeans_silhouette_analysis.py
Remove trailing spaces
MAINT fix typo pyagm -> pygamg in SkipTest
Luis Pedro Coelho (2):
DOC Fix n_jobs documentation
ENH Add cross_val_predict function
Lukas Michelbacher (3):
Fix typo
Add dependence to max_features to docstring
Add newline before bullets
MLG (1):
Quick fix on grid_scores, updating final value with len(cv)
Manoj Kumar (185):
Link my name to my wordpress blog
ENH: Return attribute n_iter_ for linear models dependent on the enet solver
TST: Added test to check that warm model converges faster than a cold one
TST: Better to test that warm_start runs only once after the prev model has converged
TST: Added test to check that higher alpha converges faster
Added return_n_iter parameter for enet (and lasso) path
FIX: Fit the data in tests
FIX: Remove raw_coef_ attribute that eats up memory
ENH: Added n_iter_ parameters across all iterative solvers
FIX: Preserve public API by introducing return_n_iter param
TST: Test that after fitting n_iter is greater than 1
Added n_iter attribute to OMP (and CV)
MAINT: Fixed a few docstrings and cosmits
COSMIT
Replaced Label Encoder with Label Binarizer
Replaced helper function _phi by special.expit
Do away with intercept helper functions
Refactor fit_intercept case
FIX: Fixed hessian value for intercept
DOC: Add docs for helper functions
ENH: LogisticRegressionCV can now handle sparse matrices
TST: Add tests to explicitly check hessian, loss and gradient for fit_intercept
More docs and tests for LogisticRegressionCV
FIX: Doctests
TST: Improved tests
ENH: Added one-vs-all fit in case of multi-class data
ENH: Added refit parameter
TST: Tests to verify OvA behavior
FIX: PEP8 and other cosmits
Made the following changes
ENH: Logistic Regression now supports newton-cg and lbfgs
ENH: Weighted logistic regression for lbfgs and newton-cg
Changed copy default from True to False, updated docstring for sample_weights
DOC: Minor changes
ENH: Added warnings for convergence, added support for l1 penalty if solver is liblinear
FIX: Changed tolerance of newton-cg to be compliant with that of lbfgs
COSMIT: Utils imports are together
FIX: Class weights are computed for each OvA
FIX: Liblinear solver for LogisticRegressionCV now works for class_weight==auto
FIX: Add DataConversionWarning
FIX: Changes due to recent refactoring of the check functions
MAINT: Improve documentation and coverage
Update whats new!
FIX: Fixes for the cross_validation failure
FIX: Increase testing accuracy
Improve docstring of the LogRegCV model
ENH Stochastic Coordinate Descent for ElasticNet & Lasso
Max iter param in Liblinear is now softcoded
ENH: Return n_iter_ from liblinear and print convergence warnings
TST: Added test to check ConvergenceWarning
Fix heisenbug due to addition of max_iter param in LSVC
FIX: Incomplete download of 20newsgroup_dataset
FIX: Memory crashes for _alpha_grid in ENetCV
ENH: Add support for class weights
FIX: Change alpha to 1./c
ENH: Merge MultinomialLR into LR with multi_class='multinomial'
ENH Added multinomial logreg to plot_classification_probability.py
ENH: Added multinomial option to LogisticRegressionCV
COSMIT: Minor doc fixes
TST: Tests for multinomial logistic regression
COSMIT: Made the following changes
DOC: Improved documentation and error messages
DOC: Made the docs for LogisticRegression clearer
FIX: PEP8 Errors and unused imports
DOC: Changed docstring style for optional arguments
DOC: Explicit instructions for Python3
MAINT: Remove BaseLibLinear for LogisticRegression
DOC: Added np.sqrt since default is 'squared=False'
ENH: Improved verbosity and denesting
FIX: Fix the imputation example
MAINT: Made the following changes
DOC: Explain prediction when decision_function is zero
OPT: Prevent iterating across n_features twice
FIX: Make sure LogRegCV with solver=liblinear works with sparse matrices
Changed default argument of precompute in ElasticNet and Lasso
TST: Added tests to test Deprecation warning
DOC: Document criterion_ attribute in LassoLarsIC
FIX: Raise warnings in f_classif a given feature is constant throughout
ENH: More descriptive error which prints the feature indices
FIX: Thresholded Nearest Centroid fails with non-encoded y
Bug fix in computing the dataset_centroid in NearestCentroid
FIX: Remove test that tests state of warnings before and after assert_warns
Merge pull request #3750 from MechCoder/bug_nearest_centroid
TST: Test that the warning registry is empty after assert_warns
TST: Test that assert_warns is reset internally
DOC: Make it explicit that assert_warns clears the warning registry
Merge pull request #3752 from MechCoder/remove_testing_test
Expose positive option in elasticnet and lasso path
TST: Add tests and document whats_new.rst
ENH: Patches Nearest Centroid for metric=manhattan for sparse and dense data
FIX: Wrap csc_row_median around the _get_median imputer function
MAINT: Move _get_median into sparsefuncs to avoid circular imports
DOC: Explain why the centroid of the manhattan metric is the median
Renamed csc_row_median to csc_median_axis_0
Warning for non-euclidean and non-manhattan metrics
Sparse matrix conversion depending on the type of metric
Update what's new.rst
Handle_unknown option to OneHotEncoder
Merge pull request #3801 from cmd-ntrf/typo_vbgmm
DOC: Make the comment slightly clearer
COSMIT: Minor opt in pairwise_distances_argmin_min
OPT: Speed improvements by avoiding repeated calls to check_array
Merge pull request #3779 from arjoly/sw-dummy-regressor
COSMIT: Use searchsorted in weighted_percentile
Update about page
Raise error when sparse matrix is supplied to predict in KNeighborsClassifier.
DOC: Typo in dict_faces.py
Minor typo in plot_dict_face_patches.py
FIX: Raise error when patch width/height is greater than image width/height
Fix BallTree and KDTree Docs
what_new entry for KD and BallTree doc fixes
ENH: Birch, first commit
Made the following changes
ENH: Made the following changes
OPT: Set check_X_y=False internally in birch
OPT: Increased the code speed by doing the following.
OPT: Made the following changes.
ENH: Added example to compare birch and minibatchkmeans
ENH: Add arbitrary clusterer in the global clustering step and test
ENH: Sparse matrix support
TST: Add tests for branching_factor
OPT: Remove unwanted calls to np.asarray for self.centroids_
ENH: Add narrative documentation
STY: Cosmits in documentation
Made the following changes
DOC: Update documentation of _CFNode
Made the following changes
Made the following changes.
Made the following changes
Prevent preprocssing for iterating over dense data
DOC: Modification to the documentation of threshold
Made the following changes.
Made the following changes:
API: Rename n_clusters to global_clusters
Changes to narrative documentation
Major changes
Rename global_clusters to n_clusters again
FIX: Remove trailing attributes in __init__
Explicit error when n_dim changes during partial_fit
Rewrite model if fit is being called after partial_fit or vice versa
Update whats_new.rst for n_iter_ attribute
Tuple unpack farthest_dist
Made the following changes
MAINT: Moved compute_label test to common tests
MRG: Fix FeatureAgglomeration docs
COSMIT: Make how to use partial_fit render properly
ENH: Use check_X_y in pairwise_distances_argmin
FIX: Raise AttributeError for fit_predict
ENH: Add metric support to neighbors_graph
Replace feature with sample
Remove optional input validation from pairwise_metrics
Merge pull request #3946 from MechCoder/doc_fix
Validation of params passed to neighbors_graphs
ENH: Allow connectivity to be a callable in AgglomerativeClutering et.al
Remove support for int and update whatsnew.rst
Use check_array and remove shape check
DOC: Improvements to children_ attribute in AgglomerativeClustering
Borrow a bit from scipy definition
ENH: Return distances option for linkage_trees
Merge return_distance tests of both linkage and ward tree
Add documentation for return_distances option
COSMIT: Removed XXX comment related to LIL matrices
FIX: Fix bug in computing full tree when n_clusters is large
DOC: Add note to whatsnew.rst
Merge pull request #3978 from MechCoder/compute_full_tree_bug
Merge pull request #3976 from MechCoder/agglomerative_playground2
TST: Test ground truth in linkage trees
Merge pull request #4000 from he7d3r/patch-1
FIX: Bug in sparse coordinate solver in lazy centering
MAINT: Remove deprecation warnings in enet_path and lasso_path
Merge pull request #4061 from ragv/pep8_cleanup
FIX: Regression in NearestCentroids
FIX/ENH: Fix kneighbors_graph/kneighbors and allow X=None
FIX: Handle and test case where n_duplicates > n_neighbors
Add docstring and test for kneighbors
ENH/FIX: Patch radius_neighbors to follow kneighbors convention
TST: Refactor tests for k and radius neighbors
ENH: Add include_self param to graph variants for backward compat
DOC: Add what's new entry
DOC: Make docs clearer
Merge pull request #4046 from MechCoder/allow_X_none
Merge pull request #4222 from saketkc/benchmark_fixes
Merge pull request #4226 from ragv/fix_4224
FIX: Fix n_components in AgglomerativeClustering and friends
Merge pull request #4277 from Celeo/contributing-request-typo
Manuel (1):
Rename k to n_clusters in docs
Mario Michael Krell (1):
DOC typos and consistency in SVM docstrings
Martin (2):
added test for transformed scatter matrix
DOC Updated what's new
Martin Billinger (5):
Added test for orthogonal LDA transform.
Fixed LDA transform.
Improved precision of the LDA orthogonality test.
forgot to remove an import
misc
Mateusz Susik (1):
FIX Support for negative values of n_jobs
Mathieu Blondel (21):
Merge pull request #3365 from fabianp/master
Fix wrong attribute doc in BaggingRegressor.
Merge pull request #3667 from ajschumacher/patch-6
Wording.
Fix typo.
Merge pull request #3713 from akshayah3/Boost
Fix sparse_cg solver when max_iter is specified.
Add random forests to document classification example.
Typo
Merge pull request #4047 from agramfort/faster_sparse_enet_dual_gap
Update lightning information.
Update seqlearn information.
Remove singular matrix warning.
Remove unused imports.
Merge pull request #4117 from chyikwei/fix_import_err
Kernelized ridge regression -> Kernel ridge regression
Fix error in CCA docstring.
Handle sample_weight upfront.
Simplify rescaling.
Use relative imports.
Add what's new entry.
Matt Pico (1):
DOC on negation of loss functions used as scoring parameters
Matt Terry (1):
FeatureUnion example using a heterogeneous datasource.
Matteo Visconti dOC (18):
FIX: ward_tree now returns children in the same order for both
Add forgotten newline
Cosmetic change in test
Minor changes: float conversion and rename rnd into rng
Merge remote-tracking branch 'upstream/master' into wtree_children
ward_tree returns heights (distances) between the clusters
ENH: `ward_tree` returns distances
changed xrange to range for compatibility with Python 3
TST, test ward_tree_distance on known dataset
FIX: structured and unstructured ward_tree return children sorted in the
Cosmetic change in test_hierarchical, better documentation
Add ENH description in whats_new.rst
Add better description of distances, cosmetic changes
Fix git rebase problems
Add forgotten IFs
Clean code
Clean code
DOC: fix correct shape of children and distance
Matti Lyra (1):
Fixed issue 3815. Discrete AdaBoostClassifier now fails early if the base classifier if worse than random.
Max Linke (2):
MAINT clean up flake8 complaints in k-means
ENH: Precompute distances only if overhead is below 100MB
Mehdi Cherti (1):
Add predict_log_proba method to GradientBoostingClassifier
Michael Bommarito (2):
Adding unit test to cover ties/duplicate x values in Isotonic Regression re: issue #4184
Adding fix for issue #4297, isotonic infinite loop
Michael Eickenberg (4):
FIX Exact inverse transform for whitened PCA
TST test whitened inverse
ENH components of unit length and whitening done by scaling with sqrt of explained variance
DOC What's new
Michal Romaniuk (1):
Enable grid search with classifiers that may throw an error on individual fits.
Nelle Varoquaux (1):
DOC updated installation documentation
Nicola Montecchio (1):
unused variables are redefined later
Nikolay Mayorov (5):
FIX Implemented correct handling of multilabel y in cross_val_score
FIX bug with set_params on minkowski 'p'.
Fixed a bug in RFECV when step != 1
Explicit encoding for opened files in gen_rst.py
Support Python 2 and 3
Noel Dawe (13):
tree: add min_weight_fraction_leaf
forest: add min_weight_fraction_leaf
min_weight_fraction_leaf and min_samples_leaf: test both best-first and depth-first tree building
gbrt: min_weight_fraction_leaf value test
tree: use min_weight_leaf internally while exposing min_weight_fraction_leaf
utils.testing: add assert_greater_equal and assert_less_equal
forest: test min_samples_leaf and min_weight_fraction_leaf
min_weight_fraction: comments and pep8
test assert_greater_equal and assert_less_equal
min_weight_fraction_leaf: narrative docs
min_weight_fraction_leaf: narrative doc update
scorer: add sample_weight support
plot_adaboost_twoclass.py: minor improvements
Okal Billy (2):
bug fix for issue #3526
bug fix for t-SNE (issue #3526) with new inputs
Olivier Grisel (176):
MAINT skip tests that require large datadownload under travis
MAINT add docstring to explain the motivation of the fixture
Typo
Merge pull request #3355 from larsmans/optipng
TST fix precision failure on windows
MAINT skip joblib multiprocessing tests on travis
Merge pull request #3361 from ogrisel/travis-no-multiprocessing-joblib
FIX assert_array_almost_equal for windows tests
FIX multilabel deprecation warning in RidgeClassifierCV
FIX #3372: unstable input check test for RANSACRegressor
MAINT ensure that examples figures are displayed in the correct order
DOC Refreshed the documentation to install scikit-learn
MAINT: skip some unstable transformers test under Win 32
MAINT configure Windows CI with appveyor.com
DOC various typos and fixes for the installation doc
DOC phrasing
MAINT master to 0.16-git
DOC whats_new.rst format fix
MAINT .mailmap update
MAINT whats_new: contributors for 0.15
DOC Copy and paste error, thanks @larsmans.
DOC update news on the homepage
DOC broken formating for the People list of 0.15
MAINT update doc version navigation
Merge pull request #3377 from ldirer/hashing_fix3356
MAINT update doc version links in support.rst
MAINT point to the configuration of the docbuilder server
MAINT temporary fix to handle cython renaming in 0.15
MAINT 32 bit unstable tests are unstable on all OSs
MAINT typo in Makefile
Merge pull request #3349 from MechCoder/return_niter
TST non-regression test for CV on text pipelines
Merge pull request #3400 from arjoly/forest-test-oob
Merge pull request #3396 from amueller/less_loud_randomized_lasso
Merge pull request #3397 from arjoly/tree-factor-rand
MAINT faster test_weight_boosting
TST speedup test_spectral_biclustering
TST faster check_transformer_pickle by fixing n_iter
TST speedup test_permutation_score
TST remove joblib tests from sklearn
Merge pull request #3403 from ogrisel/speedup-tests
Merge pull request #3460 from jnothman/fix_nn_example_doc
Merge pull request #2777 from jnothman/doc_linkcode
FIX better RandomizedPCA sparse deprecation
Merge pull request #3470 from ogrisel/fix-sparse-randomized-pca-deprecation
MAINT disable verbose tests on travis
MAINT Move the build_ext --inplace call to install.sh
Merge pull request #3441 from ogrisel/travis-no-verbose-tests
MAINT More robust windows installation script
MAINT move skip for unstable 32bit to _check_transformer
FIX unstable test on 32 bit windows
FIX numerically unstable test_logistic.py
DOC whats_new.rst entry for warm_start forests
Merge pull request #3454 from ldirer/confusion_matrix_example
TST stabilize check_transformer_n_iter
FIX revert changes introduced by mistake in previous commit
Merge pull request #3492 from furukama/patch-2
COSMIT smipler & more robust check in test_kfold_valueerrors
MAINT fix prng in test_f_oneway_ints
DOC whats_new.rst: missing backported fixes for 0.15.1
FIX #3485: class_weight='auto' on SGDClassifier
FIX wording in whats new in 0.15.1
ENH removed manual code for parallel task batching forests
ENH: parallel feature importances for forests
FIX #3566: redefine isfinite alias in sklearn
Merge pull request #3612 from calvingiles/fix-resource-warnings
ENH cosmetic reorg of the confusion matrix example
FIX: divide each row of the CM by the true support
FIX bump up the miniconda installation script
Merge pull request #3619 from ogrisel/fix-miniconda
FIX define CC and CXX for travis
FIX Windows CI: use prebuilt numpy / scipy
FIX heisenfailure in test_lasso_lars_path_length
Merge pull request #3629 from ogrisel/fix-unstable-lars-test
FIX heisenfailure on 32 bit python + speedup
FIX finfo.eps should be used instead of tiny for equality checks
ENH upload Windows wheels to rackspace
DOC fixes in 0.15.2
FIX more tolerant early stopping for least angle
MAINT CI: reflect CLI change in wheelhouse-uploader tool
FIX #3370: better lars alpha path inequality checks for 32 bit support
ENH comment on drop for good not being triggered in the tests
Merge pull request #3224 from pprett/gbrt-sample-weight
Merge pull request #3285 from kastnerkyle/incremental_pca
MAINT use Python versions provided by AppVeyor CI
FIX call srand whenever random_seed >= 0 in libsvm
MAINT missing numpy / scipy for the Python 3.3 build
MAINT remove integer warnings under Python 3 + numpy 1.8+
FIX warning check in test_affinities
FIX #3503: use linalg.pinv to better deal with singular input data
MAINT CI: collect rackspace DNS info
MAINT try to avoid HTTP timeouts to rackspace
FIX #2986: ZeroDivisionError in LinearRegression on sparse data
Merge pull request #3853 from ogrisel/backport-lsqr-fix
FIX make assert_raises_regex backport for 2.6 consistent with 2.7+
ENH use specific warning class for RP
Merge pull request #3860 from ogrisel/custom-rp-warning
Merge pull request #2949 from FlorianWilhelm/theilsen
Merge pull request #3173 from arjoly/sparse-tree
Merge pull request #3870 from FlorianWilhelm/whats_new_theilsen
DOC update whats_new.rst (sparse data for trees)
Merge pull request #3826 from amueller/kde_y_none
Merge pull request #3871 from anntzer/fast-affinity-propagation
Merge pull request #3858 from banilo/km_init
MAINT bump joblib to 0.8.4
DOC update whats_new.rst for AffinityPropagation optim
FIX clip SGD gradient on linear models for stability
TST update numerical overflow (non-regression) tests
DOC update whats_new.rst for SGD stability
DOC target URLs on SGD loss / penalty figures
Merge pull request #3882 from MechCoder/doc_tree
MAINT enable verbose output to track random failure on appveyor
Merge pull request #3902 from JeanKossaifi/doc_enhancement
Merge pull request #3915 from ThomasDelteil/fix_css_examples
ENH do not use plt.cm.jet in JL bound example
ENH improve LSHF scalability example
DOC: explain the importance of iid index and queries in bench script
ENH: ensure that max hash size and hash dtype are consistent
ENH more LSH scalability example and doc improvements
MAINT dev version to follow PEP440
MAINT more informative assertion failure in test_common
MAINT make it possible to use wheelhouse-uploader
Merge pull request #4013 from amueller/deprecated_0.16
Merge pull request #3991 from jnothman/lshforest_improvements
Merge pull request #4030 from amueller/add_pipeline_docstrings
Merge pull request #4031 from amueller/check_scoring_fixes
ENH better RBF parameters heat map
Merge pull request #4095 from jjhelmus/bug_local_import
Merge pull request #4058 from amueller/cleanup_common_tests
Merge pull request #4018 from ragv/fix_for_nan_scale
STYLE cosmetics / unused import
FIX #4059: explicit warning for strict selectors
Merge pull request #4186 from xbhsu/non-inheriting_estimator_tests
Merge pull request #4064 from amueller/pipeline_scoring_y_none
TST more strict selectors with empty support
MAINT enable travis notifications in gitter
MAINT enable appveyor notifications in gitter
FIX tests in gmm broken by previous rebase
Merge pull request #4233 from amueller/minor_doc_improvement_ensemble
FIX buggy idiom found by pyflakes
FIX regularized covariance on 1D data
FIX stricter validation for non-empty input data
FIX broken test under 64 bit Python 2 / Windows
Merge pull request #4251 from ogrisel/fix-shape-format
Merge pull request #4266 from lesteve/python3-fix-plot-kmeans-silhouette-analysis
MAINT mark _sigmoid_calibration private
ENH ensure that a warning is raised when sample_weight is not supported
DOC update contributors for the calibration features
FIX DOC TST: alpha is an uppper bound on FDR
Merge pull request #4057 from amueller/dtype_object_conversion
DOC cosmetics in test docstring
DOC document GMM fix in whats_new.rst
Merge pull request #4284 from hbredin/dgpmm_convergence
Merge pull request #4261 from ragv/svm_scale_c_plot
DOC update whats_new for deterministic spectral_embedding
FIX test for consistent handling on empty input data
DOC whats_new.rst for BernoulliNB fix
TST check that constructor params are not mutated by fit
DOC whats_new.rst for radius_neighbors boundary handling
Merge pull request #4318 from amueller/skip_omp_cv_on_travis
Merge pull request #4313 from vortex-ape/spectral_clustering
Merge pull request #4192 from jnothman/binary_iff_binary
Merge pull request #4307 from amueller/more_quite_testing
Merge pull request #4326 from Barmaley-exe/svm-dual-coef-fix
DOC fix broken image link
DOC add some missing API links in whats_new.rst
STYLE trailing spaces
ENH no need for tie breaking jitter in calibration
ENH improve docstrings and test for radius_neighbors models
TST boundary handling in LSHForest.radius_neighbors
Merge pull request #4317 from ogrisel/fix-radius-queries
Merge pull request #4028 from ogrisel/wheelhouse-uploader
DOC deprecate random_state for DBSCAN
Merge pull request #4302 from amueller/isotonic_regression_duplicate_fixes
MAINT use canonical PEP-440 dev version consistently
FIX #4358: make weighted scorer test deterministic
Paolo Puggioni (1):
add crossvalidation
Peter Prettenhofer (23):
fix: i might be undefined
make IsotonicRegression pickleable
import pickle
Sample weights for gradient boosting
more tests
add exponential loss to narrative documentation
cosmit: doc
bincount supports weights
more elegant PriorProbabilityEstimator
adressed Olivier's feedback
update tests for new exception types
sample weight support for robust regression via weighted percentile algo
fix: consider sample_weights in robost init estimator and negative_gradient
fix: add **kwargs to Multinomial loss' negative_gradient
more tests for weighted percentile
cosmit
add GBRT sample weights to whats new
fix: sample_weight is None in Huber.deviance
cosmit arr -> array
non-uniform weight test case for both reg and clf and all losses
add probability output for exponential loss
we can produce probabilistic outputs using exponential loss
Merge pull request #3669 from pprett/fix-isotonic-pickling
Pratap Vardhan (2):
MANIT: pep8ize to an extent cluster module
MANIT: Revert to py3 print version
Rafael Carrascosa (1):
DOC: added Machinalis testimonial
Raghav R V (46):
Fixes #3644
Added NRT for return_path propagation check
DOC added note to warn users not to __init__ params with trailing _
ENH FIX Raise ValueError when the target does not match with classes arg
TST Added NRT to test if ValueError is raised during target mismatch.
PEP8 Fix E112 and E113 errors
COSMIT Remove trailing : in comment to facilitate pep8 autoindentation
PEP8 Fix E101, E111 errors and W191, W293, W293 and W391 warnings.
COSMIT / PEP8 Limit line length to 79
MAINT Add new function to check if an estimator is fitted
MAINT Make uniform the error raised for not fitted condition
MAINT Remove the _check_fitted test.
FIX the try except block to recognize NotFittedError
TST Add tests to assert if NotFittedError is raised appropriately
FIX support_ parameter should be tested, not the coef_
FIX Refactor _is_fitted method into check_is_fitted.
FIX Make error message more explicit in fit_inverse_transform
DOC/MAINT Update whats new entry for NotFittedError
FIX scale - Raise ValueError when input contains non finite values.
TST Add test to check if warning is raised for integer dtype-ed inputs.
TST Add test to check if ValueError is raised if input contains NaN
FIX Remove unused score_overrides_loss param from check_scoring
FIX various mismatch between docstring and signature params
DOC Add silhouette analysis plot for KMeans
DOC ENH Fill the silhouette with the corresponding cluster color
DOC ENH Remove the vertical x = 0 line and plot avg silhouette score line
FIX Use slinear interpolation for isotonic regression
MAINT Remove temporary fix #3995 in view of the change to slinear.
DOC ENH Simplify the example code; Add plots for n_clusters = 3 and 5
DOC Added what's new entry for silhouette plot example
DOC Correct what's new entry of Jaccard Similarity Score
MAINT threshold -> threshold_ with deprecation warning.
TST Removal or modification of stop_words_ should not affect transform.
ENH decision_function now returns #votes and scaled scores.
PEP8 Clean up minor PEP8 issues.
FIX Return np.zeros(n_alphas) when the targets are uniform
TST Add enet test for the case of uniform targets
FIX Use resolution instead of min from np.finfo(float)
TST Check for equality in alphas_ when uniform targets are used.
FIX Add check_symmetric to __all__
FIX loss, penalty --> penalty, loss
EXAMPLE/FIX l2 --> squared_hinge; L1/L2 --> l1/l2
FIX/MAINT Deprecate and support upper case values for loss/penalty
COSMIT Flake8 fixes for svm/base.py
FIX Make LinearSVC/LinearSVR support the deprecated l1/L1, l2/L2
TST test deprecation warning and support for uppercase
Rahiel Kasim (1):
Repeated word: 'the the' -> 'the'
Randall Mason (1):
Spelling in heading newgroups -> newsgroups
Robert Bradshaw (1):
Lift division out of loop in _isotonic_regression.
Saket Choudhary (2):
DOC: Fix is->if
FIX precompute_gram->precompute as in #2224
Sam Nicholls (1):
FIX Verbose message in RFE to count from 1, not 0
Samuel Charron (1):
OPTIM Lower memory usage of spectral_embedding.
Saurabh Jha (2):
Implements Multiclass Hinge loss
Sample weight support for hinge loss
Stefan Otte (1):
DOC remove unnecessary backticks in CONTRIBUTING.
Stefan van der Walt (1):
ENH Speed up and simplify cartesian product
Steve Tjoa (1):
fixed typo: diriclet -> dirichlet
Sylvain Zimmer (2):
Fix typo in SVM docstrings
Another doc typo fix
Thomas Delteil (1):
reverted css to current stable, fixes the display bug
Thomas Unterthiner (7):
MAINT explicit float64 in sparsefuncs_fast
ENH Add 'axis' argument to sparsefuncs.mean_variance_axis
ENH improved CSC matrix handling in VarianceThreshold
FIX consensus score on non-square similarity matrices
ENH add testcase for issue 2445
ENH: cross-validate over predefined splits
COSMIT pep8 cleanups
Tomas Kazmar (1):
FIX MinMaxScaler behavior on 1D inputs.
Tzu-Ming Kuo (1):
MAINT liblinear class label logic (disabled)
Vinayak Mehta (3):
explicitly mentioned include_self
explicitly passed by name
Added Random Kitchen Sink reference
Vincent Dubourg (2):
FIX: random_start feature in GaussianProcess
FIX: gp_diabetes_dataset examples (theta_ attribute)
Vjacheslav Murashkin (1):
FIX handle selection in feature names for example
Vlad Niculae (9):
TYPO remove mutli (did you mean Muttley?)
Turn useless line of code into descriptive comment
DOC fix docstrings; add @hamsal to authors
COSMIT Use explicit if/else in scorer
TST default scorers with sample_weight
DOC Update What's New
FIX classes name in OvR
FIX set vectorizer vocabulary outside of init
Deprecate vectorizer fixed_vocabulary attribute
Will Lamond (2):
FIX allow ndim>2 in shuffle
Fixes ovr in the binary classifier case, and adds support for lists of feature arrays in the multiclass case.
Wu Jiang (1):
Fix a typo in OneHotEncoder's docstring
Xinfan Meng (1):
DOC Fix broken class references
Yan Yi (2):
FIX inconsistent boundary handling
TST add test for radius_neighbors boundary handling
Yaroslav Halchenko (9):
Merge tag '0.16b1' into releases
Merge branch 'releases' into dfsg
Merge branch 'dfsg' into debian
Merge remote-tracking branch 'origin/0.16.X' into debian
updated patches
mark autoreconf for cython ge 0.20.1
Merge remote-tracking branch 'origin/0.16.X' into dfsg
Merge branch 'dfsg' into debian
2 new debian/patches
Yu-Chin (4):
COSMIT tabs/spaces in Liblinear
MAINT variable initialization in Liblinear
MAINT use size_t in Liblinear
MAINT changes to (disabled) Liblinear code, from upstream
abhishek thakur (1):
fix svm.fit in feature_stacker.py
akshayah3 (15):
Added Deprecation warning
Small PEP8 fix, n_jobs attribute is compatible with both fit method and the constructor
Fixed a failing test in the doc
Added tests
Added seperate tests
Refactored some logic code
PEP8 fixes
Added make_fit_parameter method
dded tests to test_boosting
Added _check_sample_weight
Shifted the _check_sample_weight into BaseWeightBoosting class
Raise exception for sparse inputs in the case of svd solver
Documentation of max depth parameter
Sparse inputs for fit_params
Added helper function
banilo (2):
added n_init check for k_means()
FIX: typo in Pipeline error
bhsu (2):
non-inheriting estimator added for rfe, cross validation, and pipeline tests
PEP8 compliance
cgohlke (1):
MAINT Include binary_tree.pxi in source distribution
chebee7i (1):
DOC: Reword docstring and deprecation warning for include_self.
danfrankj (1):
Correct documentation for TfidfVectorizer
floydsoft (1):
fix the link of Out-of-core_algorithm
gpassino (1):
DOC fix kernels documentation inconsistencies
gwulfs (1):
PyMC
isms (1):
DOC: Fix LabelBinarizer docstring to rst format bug
jdcaballero (1):
Update plot_outlier_detection.py
leepei (2):
MAINT trailing whitespace in linear.cpp
Let nr_fold = l when nr_fold > l in CV
maheshakya (56):
ENH: Implemented LSH forest
TST tests for LSHForest
ENH Added radius_neighbors method to LSHForest.
MAINT refactor LSH forest code
Updated docstrings and tests.
Used bisect in _find_matching_indices method.
Removed cached hash and replaced with numpy packbits.
Updated lsh_forest and test_lsh_forest.
Updated docstrings for tests.
DOC Added example for LSHForest.
FIX Updated tests. renamed lsh_forest to approximate.
DOC Updated the example.
COSMIT pep8 LSH code
DOC Updated the document of neighbors module.
FIX Completed tests for missed lines.
FIX Updated examples
Removed redundant ran_seed from GaussianRandomProjection Initialization.
Updated Approximate nearest neighbors in neighbors.rst
Inlined generate_hash into create_tree in LSHForest.
Updated test_approximate.
Re-written Approximate nearest neighbors documentation.
FIX Moved random_state to fit method to pass seed to create_tree.
FIX Updated plots in example.
DOC Included information about distance measure and LSH method in the documentation.
Added test to check randomness of hash functions.
FIX Added integer as random_state to test_hash_functions.
Updated example to avoid redundants.
Example splitted into hyperparameters and scalability.
FIX Fixed accuracy error in scalability example.
DOC Added new section-Mathematical description of LSH.
Created benchmark for approximate nearest neighbors.
Added scalability plots into documentation.
FIX Modified tests to handle cosine distances instead or euclidean.
Updated example with cosine measures.
Modified plots in hyper-parameters and scalability plots.
FIX Added user warnings for insufficient candidates.
Changed parameter descriptions of LSHForest.
FIX LSH: Fill up candidates uniformly from remaining indices
Added KNeighborsMixin, RadiusNeighborsMixin.
Added tests for graph methods.
FIX Fixed benchmark.
ENH Added error bars to scalability example plots.
FIX Fixed ordering of returned neighbors.
DOC Modified documentation of approximate neighbors.
FIX Fixed example.
ENH Modified hyperparameters example to calculate stadard deviations of accuracies.
ENH Modified approximate neighbors doc string.
ENH Changed doc string and comments in test_distance function.
ENH Used GaussianRandomProjectionHash for handling hashing.
ENH Changed description of hyper-parameters example.
ENH Inlined _create_tree method.
DOC fix LSH forest docstrings
ENH Modified kneighbors and radius_neighbors methods.
FIX Fixed partial_fit.
TST improve tests for approximate neighbors.
FIX Modified assert_warns_message to ignore order of warnings.
mr.Shu (1):
DOC: COSMIT: Naive Bayes long url + PEP8 cleanup
pvnguyen (1):
Update outlier_detection.rst
queqichao (4):
+ Fix the bug of calling len when using sparse matrix for fit_params in cross_validation.py.
+ Fix the bug of calling len when using sparse matrix for fit_params in cross_validation.py.
Merge branch 'fix_bug_in_cross_validation_when_using_sparse_matrix_for_fit_params' of https://github.com/queqichao/scikit-learn into fix_bug_in_cross_validation_when_using_sparse_matrix_for_fit_params
+ Fix a typo.
sethdandridge (1):
Fixed error message typo
snuderl (1):
Small typo.
swu (16):
Adding newton-cg solver for multinomial case.
Add comments to _multinomial_loss_grad_hess function.
Fixed pep8 errors. Added docstring for _multinomial_loss.
Fixing docstring on _multinomial_loss.
Fixing more pep8 errors.
Merge remote-tracking branch 'upstream/master' into multinomial_newtoncg
Removed some unnecessary lines and reformatted for consistency in _multinomial_loss and _multinomial_loss_grad_hess.
Modified parameter description in docstring of
Adding test cases for multinomial LogisticRegression using newton-cg
Modified hessp function in _multinomial_loss_grad_hess to compute r_yhat
Update whats_new.rst. Add contact information to logistic.py.
Modified doc/modules/linear_model.rst to incorporate the newton-cg solver
Minor doc and formatting changes.
Reformatted docstrings in multinomial functions. Refactored multinomial
Wrap _multinomial_loss_grad in lambda function for passing into lbfgs.
Modify shape check on coef matrix in _logistic_regression_path.
tejesh95 (1):
Update gaussian_process.py
terrycojones (1):
Added missing space to exception message. Simplified (trivially) code and corrected message. Updated tests.
trevorstephens (19):
GradientBoostingClassifier docstring incorrectly specified default for max_features as "auto" when it is None.
various docstring fixes for web docs
add support for class_weights
add multioutput support & bootstrap auto mode
expanded class_weight dimension fix
add class_weight to trees, expand tests & minor refactor
parameter validation checks & tests for errors
Y-org rename & whats_new update
Merge branch 'master' into rf-class_weight
rename vars & copy sample_weight
Merge branch 'master' into rf-class_weight
rename cw option to subsample & refactor its implementation
fix nb partial_fit w class_prior #3186
fix rst table layout
add compute_sample_weight util
refactor forests & trees class_weight calc
use safe in1d, use compute_sample_weight in ridge
remove inplace multiplication in ridge
add test for None class_weights
tttthomasssss (1):
fix for issue #4051; replaced X = np.asarray(X) with check_array(...) method
ugurcaliskan (2):
Update plot_rfe_with_cross_validation.py
Update plot_multilabel.py
ugurthemaster (2):
Update plot_tree_regression.py
COSMIT Move imports in example
wadawson (1):
Corrected two bugs related to 'tied' covariance_type in mixture.GMM(), added test, closes #4036
x0l (6):
FIX error handling/memory deallocation in Liblinear wrapper
Fix a bug occurring when n_samples in a class < n_features
warning + log of prod -> sum of log
warning -> error
tests
ignore warnings, change docstring
Óscar Nájera (10):
A short cleanup
recovering deleted variable
gen_rst informs file with missing documentation
fix tab error
gen_rst to use css tooltip instruction
remove gallery css chunk from theme css
Remove JS from examples gallery
Examples gallery css file
fix gen_rst file extension
facecolor/edgecolor are passed arguments to savefig if they
-----------------------------------------------------------------------
No new revisions were added by this update.
--
Alioth's /usr/local/bin/git-commit-notice on /srv/git.debian.org/git/debian-science/packages/scikit-learn.git
More information about the debian-science-commits
mailing list