[scikit-learn] annotated tag debian/0.15.0_b1-1 created (now a8290de)
Andreas Tille
tille at debian.org
Wed Dec 28 13:11:04 UTC 2016
This is an automated email from the git hooks/post-receive script.
tille pushed a change to annotated tag debian/0.15.0_b1-1
in repository scikit-learn.
at a8290de (tag)
tagging ba1914c75d48510f8664e25a296b5fdb170592a2 (commit)
replaces debian/0.14.1-3
tagged by Yaroslav Halchenko
on Fri Jun 6 12:05:59 2014 -0400
- Log -----------------------------------------------------------------
scikit-learn Debian release 0.15.0~b1-1
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1
iEYEABECAAYFAlOR5ucACgkQjRFFY3XAJMjVGwCfdDWi4zyZCjgAVu0FVNh6AFAp
1IMAn0HNSl3C+ct1Uxz2As4Cc9mz/jBE
=LWLV
-----END PGP SIGNATURE-----
Aaron Schumacher (5):
target is response, not explanatory
typo: "fot" -> "for"
typo: "requiered" -> "required"
DOC: typo: 1e-2 is 0.01 not 100
DOC remove import scikits.learn from tutorial
Alexander Fabisch (83):
FIX Do one run with MiniBatchKMeans and explicit centers
Add learning curve
Refactor cv code
Clean up
Refactor RFE and add _check_scorable
FIX typo in docstring
Merge `fit_grid_point` into `_cross_val_score`
Return time
Move set_params back to fit_grid_point
Log score and time in 'cross_val_score'
check_scorable returns scorer
Clean up
Replace '_fit_estimator' by '_cross_val_score'
Fix PEP8, style and documentation
Remove wrong variable names
Remove helper function '_fit'
Remove 'fit_grid_point' from 'BaseSearchCV'
Check substrings of error messages
Rename '_split' to '_split_with_kernel'
_passthrough_scorer is a function
Remove '_deprecate_loss_and_score_funcs'
Check error message
Use assert_raises_regexp to check error messages
Add prototype for validation curve
Add documentation and tests
Improve test coverage
Improve interface and documentation
Mocks inherit from BaseEstimator
Correct docstring
Add narrative documentation
Improve documentation
Fix link to validation curve
Add example with polynomial regression
Correct and improve documentation
Improve phrasing
Rephrase sentence
Simplify first part of the documentation
Fix typo
Return scores of all folds
Fix documentation
Improve test coverage
Matching colors
MAINT Rename over/underfitting example
Add t-SNE
Generalize with average of nearest neighbors
Shorten documentation of perplexity
Compare t-SNE with other manifold learners
Simplify optimization
Refactoring
Add first test
Test gradient descent
Test binary search
Add t-SNE to other examples
Add more tests
Use regular expressions
100% test coverage
Remove example with digits dataset
Rename dist to affinities
Modify learning schedule
Fix tests
Explain example
Rename attributes
Document attributes
PEP8
Update documentation
Do not stop too early
Describe how to set the learning rate
Remove generalization
Add section in narrative documentation
Adress Gael's comments
Do not use global random number generator
Replace squareform(pdist(*, "sqeuclidean"))
Adress Olivier's comments
Mention Barnes-Hut-SNE and fast optimization
Integrate PCA initialization
Use PCA initialization in examples
Fix docstring
Use euclidean_distances in original space
Mention TruncatedSVD and clean up (PEP8, Pyflakes)
Affinity must be 'precomputed' or 'euclidean'
Rename arguments
Allow sparse data
Correct examples
Alexander Measure (1):
Update naive_bayes.rst
Alexandre Gramfort (108):
Merged pull request #5 from larsmans/master.
Merge pull request #2290 from dengemann/more_ica_improvements
FIX : backport tanh out param in old numpy
revert the tanh fix
ENH : simplify ProabilisticPCA covariance_ computation + misc in pca.py
ENH : avoid extra alloc in _infer_dimension_ in pca.py
FIX : self.component in PCA was changed by fit when passed mle or explained variance
ENH : avoid extra allocation in FactorAnalysis
ENH : use math.log and not np.log for scalars
doc fix
misc
FIX : broken fit_transform
MISC : pep8 and pyflakes on test_isotonic.py
Merge pull request #2412 from rolisz/patch-2
FIX : fix randomized SVD in FA
Merge pull request #2406 from dengemann/tinker_fa
ENH : add score to PCA from ProbabilisticPCA
API : deprecating ProbabilisticPCA
misc
typo
ENH : make PCA scoring work with n_features is big. Avoid covariance computation in PCA.fit
DOC : remove reference to ProbabilisticPCA in doc
update what's new
FIX : do not store precision_ attribute in score method
ENH : add get_precision method with matrix inversion lemma to PCA for faster scoring
ENH : add tests for get_precision in PCA + some fixes for corner cases
ENH : add get_precision method with matrix inverse lemma to FactorAnalysis + use precision in score
pep8
ENH : add PCA+FA model selection example
pep8 + misc
ENH : add score samples to PCA
API : add score_samples in FactorAnalysis + update example
better wording
add ref to online pdf
FIX : fix ProbabilisticPCA.score with homoscedastic=True
for loop in tests of get_covariance + get_precision
typo
typo
wording
simplify FA tests after rebase
pimp example
show example in narrative doc
avoid zip + topic in doc
FIX : get rid of coef_init in Lasso in dict learning
FIX : cleanup coef_init in coordinate descent
FIX : warm_start in coordinate_descent (was ignored)
FIX : set deprecation to 0.16 in coordinate descent
API : set return_models to False by default in path functions in coordinate descent
FIX : get rid of convergence warning in test_coordinate_descent
ENH : Gaussian process for arbitrary-dimensional output spaces by @JohntheBear
ENH : simplify input checking in GP
update what's new
Merge pull request #2497 from dengemann/extend_fast_dot
ENH : better shrinkage range for ShrunkCovariance
s/n_features/n_samples in doc/datasets/index.rst
Merge pull request #2576 from ankit-maverick/issue2560
Merge pull request #2583 from amueller/doc_rbf_parameters
Merge pull request #2624 from dengemann/fix_dot
Merge pull request #2628 from dengemann/improve_ica_example
Merge pull request #2647 from trein/patch-1
FIX : explained_variance_ratio_ in RandomizedPCA
update tests
better sparse support
pep8
update what's new
add test with sparse data
address @ogrisel's comments
Merge pull request #2669 from GaelVaroquaux/phimeca
fix f_oneway with ints
pyflakes
Merge pull request #2683 from blagarde/master
Merge pull request #2708 from likang7/patch-3
Merge pull request #2706 from likang7/patch-2
TST : speed up tests + cosmit
FIX : n_jobs was missing from LassoCV
update what's new
Merge pull request #2742 from ai8rahim/issue-2741
update what's new
clarify what's new
Merge pull request #2857 from eltermann/copyright-updated
Merge pull request #2858 from ameasure/patch-1
Merge pull request #2866 from Manoj-Kumar-S/Iss2751
Merge pull request #2884 from dsullivan7/clusterdoc
Merge pull request #2895 from kaushik94/master
Merge pull request #2920 from yoni/master
Merge pull request #2931 from GaelVaroquaux/rm_solve_triangular
Merge pull request #2945 from maheshakya/olivetti
Merge pull request #2958 from hamsal/check_arrays_n_dim
Merge pull request #2951 from Manoj-Kumar-S/speed_sparse
Merge pull request #3017 from cgohlke/patch-2
Merge pull request #3035 from ogrisel/test-sgd-stability-error
Merge pull request #3022 from ogrisel/travis-old-numpy-scipy
Merge pull request #3116 from eickenberg/ridge_wrong_solver_exception
Merge pull request #3135 from ugurthemaster/patch-1
typo
Merge pull request #3151 from rajatkhanduja/fixing_examples
Merge pull request #3152 from jdowns/lfw-conditional-import
Merge pull request #3154 from rajatkhanduja/fixing_examples
DOC : fix alpha docstring in dict learning
DOC : really fix alpha docstring in dict learning
FIX : omp default param could fail
ENH : use local seed + avoid warning
Merge pull request #3200 from ugurthemaster/patch-2
Merge pull request #3176 from agramfort/omp_default
Merge pull request #3230 from bwignall/quickfix-gaussian
Merge pull request #3231 from bwignall/quickfix-cap
Merge pull request #2822 from AlexanderFabisch/tsne
Merge pull request #3254 from laurentluce/presentations-links-fix
Alexis Metaireau (4):
Configure sphinx to be able to load extensions
Fix OptionParser import
Fix little typos in the general_concepts document.
merge with upstream
Alexis Mignon (2):
MAINT remove default values from private funcs in GMM
ENH faster sampling in GMMs
Allen Riddell (1):
DOC: Update dead link in cross_decomposition.rst
Alyssa (4):
[DOC] Updated install instructions for Arch Linux
<MRG> Doc generation works in Python 2 and 3
[MRG] Addressed comments
[MRG] Addressed comment, fixed open-ended except
Amos Waterland (4):
Explicitly invoke the Python interpreter.
Use double backticks.
Remove extraneous period.
Explain the -1 Python syntax.
Andreas Mueller (21):
DOC add link to fetch_covertype to covertype narrative docs
WEBSITE add yhat testimonial
ENH make sure all "SkipTest" calls have an error message.
don't import tests
FIX don't raise file-level SkipTests
FIX website: css hickup
DOC explanatory sentence for svm grid search example
Website: add spotify testimonials
ENH make minibatch k-means use batches in prediction and computing labels
ENH use nostests yield construct for better error reporting in common tests.
enh give yielded tests nicer names
FIX remove ``.description`` from test generators as this is not thread-safe.
ENH Define fused types so only two functions are generated.
WEBSITE FIX spotify logo!
MISC remove accidentally commited file. Whoops!
DOC make loss function in SGD consistent with subgradient. Comment by Martin Jaggi :)
MAINT remove _label attribute from SVC.
Add FAQ to docs.
removed sklearn, removed statsmodel, capitalization in SciPy
Add GPU support (there is none and will probably be none) to faq.
DOC faq formulation
Andrew Ash (1):
Typo: thepath -> the path
Andrew Clegg (2):
Workaround for andrewclegg/snake-charmer#12
Fixed typo
Andrew Tulloch (3):
[Cross Validation] Use itertools.izip consistently across Python 2/3
[tests] Use Python 3 zip/Python 2 zip consistently in tests
ENH - Improve performance of isotonic regression
Ankit Agrawal (7):
Making the results NaiveBayes example more explicit
Adding doc for bin_seeding parameter in cluster.MeanShift
Wrapping line in naive_bayes.rst and specifying bin_seeding argument as optional
Returning an array mask using np.ma.getmaskarray
Adding a test for verifying the shape of imputed matrices
Using assert_equal routine from nose
Minor doc example fix
Arnaud Joly (143):
DOC Update what's new
ENH more explicit name for auc + consistency for scorer, fix #2096
DOC put the narrative documentation of roc_curve and roc_auc_score in one place
FIX search and replace misstake
ENH reduce memory overhead of storing tree ensemble
Merge pull request #2438 from arjoly/tree-mem-overhead2
Update what's new
FIX issue #1993: passing a multilabel indicator is no more noop
Merge pull request #2521 from dmedri/master
Merge pull request #2504 from glouppe/tree-tweaks
Merge pull request #2614 from zyv/patch-1
TST add generic test for averaging
ENH test that most metrics work with one sample
ENH auc averaging for multilabel-indicator format
ENH add multilabel-indicator support for average_precision
TST perform testing on average precision + simplify invariance test system
ENH add scorer support for multi-label threshold scorer
TST scorer works with multi-output decision function
flake8
TST more clean up test_metrics
DOC narrative doc for averaging support of average_precision_score
FIX decision function scorer multilabel
DOC narrative doc for roc_auc_score multilabel-indicator
DOC not all functions support multilabel-sequence format
DOC wording
FIX doctests
DOC extend roc example
DOC update example of precision-recall curve
TST properly raise ValueError...
DOC nicer plot
TST add roc_curve and precision_recall_curve on toydata
DOC TST explain how to use common tests in test_metrics
Typo
TST clean copy paste mistake
TST remove print
TST full coverage for _average_binary_score
DOC TST typo
WIP TST more sample invariance test
WIP TST adapt test for sample based metrics
TST finish to add invariance test for sample based metrics
DOC typo
DOC wording and typo
TST add test for garbage averaging input string
TST typo
DOC explain why roc auc score is useful in multilabel classification
typo
TST more corner case tests
TST ensure that exceptions are properly raised on np 1.3
Merge pull request #2629 from arjoly/fix-auc
Update what's new
Merge pull request #2633 from josericardo/FixingTypos
Merge pull request #2643 from jnothman/bench_mutilabel_metrics
Remove deprecated zero_one and zero_one_score
flake8
FIX silence numpy 1.8 warning for using non integer
Merge pull request #2656 from arjoly/np1.8-warnings
Merge pull request #2719 from amueller/remove_description_in_common_tests
Merge pull request #2717 from Manoj-Kumar-S/test_log_loss
TST Gini is equivalent to mse in binary classification
Merge pull request #2744 from arjoly/test-impurity
OPTIM MSE criterion
OPTIM more optimisation of mse
OPTIM more optimisation of mse
ENH use memset instead of for loop
Typo
Update what's new
Merge pull request #2732 from jnothman/two-array-tree
Merge pull request #2780 from ugurthemaster/patch-1
Merge pull request #2825 from bryan-lunt/master
Uniformize max_features semantics
TST add tests with constant features
Update what's news
Remove duplicate what's new
Merge pull request #2829 from arjoly/maxfeatures-seamantics
ENH Cache features value for extra trees
Merge pull request #2897 from ugurthemaster/patch-1
WIP add n_constant_features argument
WIP add constant_features array
ENH avoid splitting on constant features for best splitter
FIX bug with BestSPlitter + ENH avoid searching for a split on constant features
ENH avoid trying to split on constant features for presort best splitter
COSMIT
COSMIT
FIX deallocate memory
FIX proper inititalization
ENH use memcopy instead of while loop
COSMIT
FIX bug with invariant
Simpler swap
pep8
ENH remove unused constant
ENH document and rename EPSILON_FLOAT
DOC improve documentation for the splitting algorihtm
DOC verbose documentation of features and constant_features
Rename EPSILON_FLOAT -> CONSTANT_FEATURE_THRESHOLD + DOC n_drawn_constants & n_found_constant
DOC more comments about features and constant_features
DOC explain invariant with constant features + DOC clarify comments
DOC more comments of the splitting algorithm
Merge pull request #2886 from maheshakya/dummy_regressor
Update what's new
Merge pull request #2875 from arjoly/et-skip-invalid-
Update what's new
Merge pull request #2971 from jyu-rmn/master
Clean tree builder interface
ENH inline sme function
ENH avoid serializing the splitter & criterion + avoid storing tree parameter in the tree structure
ENH avoid serialising the random_state in the tree
ENH clean public tree strucure
ENH for the tree structure max_depth is part of the inner structure
ENH stricter separation between tree builder, splitter and criterion
ENH add a dedicated constant for min_impurity_split
Document maximal depth of the tree
ENH stop splitting for best first tree builder if pure node + change strict to unstrict inequality
ENH stricter separation between splitter and criterion
FIX tons of flake8
DOC document max_depth of TreeBuilder
ENH re-generate c file
Add space to be consistent with _tree.pyx
DOC document internal tree class
ENH rename CONSTANT_FEATURE_THRESHOLD to FEATURE_THRESHOLD + revert some pep8 fix for readability
DOC missing words
Merge pull request #2977 from arjoly/tree-cleanup
DOC example rendering
FIX numerical stability issues on 32 bit platform
Merge pull request #3049 from glouppe/tree-bestfirst
TST define only one list of name for sample weight invariance testing
FIX missing comma in METRICS_WITHOUT_SAMPLE_WEIGHT
DOC advertize GenericUnivariateSelect in the narrative documentation and api
Merge pull request #2936 from ugurthemaster/master
FIX raise memory error if constant_feature can't be realloc
Update numpy and scipy requirement
Merge pull request #3127 from rmcgibbo/_fit_and_score-docstring
pep8 remove unused variable
DOC typo
MAINT remove redundant class hierarchy + fix api perform parameter check in fit
DOC fix example in univariate selection (fix issue #3132)
FIX testing datasets have too few feature for default of SelectKBest
TST improve parameter checking + use nose assert_ functions
DOC remove comment thanks to @jnothman
Merge pull request #3178 from MechCoder/Iss3174
TST Refactor test of sklearn/ensemble/tests/test_forest.py
TST refactor oob score testing
Update what's new with adaboost sparse input support by @hamsal
Bala Subrahmanyam Varanasi (19):
fixed visual indentation errors
fix indentation errors
use spaces in indentation
fix indentation
fix E501: line too long
fixed indentation and visual indentation errors
fixed visual indentation errors
fixed indentation and visual indentation errors
fix for 'line too long' warning
pep8 fixes
fix indentation and convert tabs to spaces
fix indentation
fix too many blank lines
fix visual indent
add expected blank line
fix line endings and visual indents
fix visual indentation
fix visual indentation
removed unused import from example
Baptiste Lagarde (5):
FIX: Typo
FIX: Typo
FIX: Typos
FIX: Typo
FIX: Typo
Brian Kearns (1):
FIX regression in CountVectorizer handling of float min_df/max_df
Brooke Osborn (21):
DOC adding description for handling of the auto parameter in neighbors
BUG fix whitespace in error messages
changing private _fit to public partial_fit
removing NWS from list of stock prices
adding test to rbm
changing partial fit to be seperate from _fit
changing test
saving the random state
making adjustments to variable names and account for sparse matrices
changing xrange to range
putting cluster on new line to solve sphinx documentation formatting issue
changing formatting of documentation
Merge remote-tracking branch 'upstream/master' into rbmbranch
changing computation of batch slices to np.array_split
pep8 corrections
adding test for csr data for partial fit
pep8-ing
removing copy() and adding tests for multiple forms of sparse matrices
removing tests for multiple sparse matrix formats
adding test for csc and csr sparse matrix
adding change to whats_new
Bryan Lunt (1):
Fixed FS1995 citation.
CJ Carey (7):
DOC: grammar and spelling fixes
BUG: avoid NaNs throwing off class probabilities
fixing warning ignore type
adding test case
conform to pep8
apply the fix closer to the source of the issue
ENH: use partial sort for kneighbors selection
Charles Earl (1):
Added docstrings for model attributes in LabelPropagation and LabelSpreading
Chen Liu (6):
fix numpy deprecationWarning('using a non-integer number ...') in svm
fix converting an array with ndim > 0 to an index DeprecationWarning
using floor division in python3
handle the case when length of sign changed coefficients is greater than 1 in LARS algorithm
Merge remote-tracking branch 'upstream/master' into fix-DeprecationWarning
handle the case where length of idx in least_angle greater than one
Chris Filo Gorgolewski (1):
DOC: Fixed roc_curve docstring
Chyi-Kwei Yau (1):
COSMIT remove duplicate key 'hamming' in METRIC_MAPPING
Craig Thompson (1):
MAINT: unreachable code removed from BernoulliNB
Daniel Nouri (6):
FIX #2372: StratifiedKFold less impact on the original order of samples.
Fix accidental doctest breakage.
Instead of linking to NB, explain the problem inside the test itself.
Avoid list, preallocate a numpy array for indices instead.
Update comment with numbers for when we run with 800 samples.
Add entry for #2372 to whats_new.rst
DanielWeitzenfeld (1):
added howaboutwe testimonial
Daniele Medri (1):
Update linear_model.rst
David Cournapeau (1):
BUG: remove remaining utf-8 characters in docstrings.
Denis (1):
FIX Move pooling_func to constructor.
Denis Engemann (89):
ENH/FIX: improve logcosh function + tests
FIX/ENH: revert changes, improve doc
FIX: 80 characters
COSMITS
COSMITS
COSMITS
ENH: address discussion
ENH: use empty, not zeros
COSMITS
ENH: add fast_dot function
employ DataConversionWarning
FIX: avoid zero-dvision warnings
FIX: fix test warning
better handling of warning in test
COSMITS
FIX: workaround for missing BLAS
ENH: return np.dot if ndim does not match
CLEANUP
FIX: remove spurious test
ENH: add fast_dot shape check
ENH: cover another cornercase + fix messages
ENH: add new Warning class, improve tests, update docs
FIX: superfluous check
ENH/DOC: add performance tips on fast_dot
COSMITS
update what's new
FIX: proper doctest
ENH/FIX: address @ogrisel comments
FIX: move inline comment to its right place
FIX: performance.rst doctest
ENH: restrict fast_dot to np < 1.8
FIX: ghost diff
ENH: update version checking
ENH: put fast dot to PCA, ICA and FA
FIX: test doc np.allclose
ENH: add randomized_svd option to FactorAnalysis
FIX: misc
misc
COMSITS+DOC
ENH: use `algorithm` keyword + expose failing test
FIX: make SVD working again + add tests + update whats new
COSMITS
ENH/FIX: misc + address discussion
COSMITS
ENH: add strong comparisons
COSMITS: address discussion
COSMITS2
ENH: add check on init + appropriate test; COSMITS
ENH: address discussion + actuall make randomized default
ENH: exposed n_iter; add missing fast_dot
ENH: improve wording + recommendation
COSMITS3
ENH: add warning instead of print
ENH: deprecate `verbose` parameter
ENH: optimze euclidean norms
FIX scipy API
... add the fix to import check
ENH: improve FA vs PCA example
COSMITS
MAINT: use stable blas getter API
FIX: unintended assignment to ValueError
STY: tuple for Gael
ENH: refactor warnings 1
ENH: add assert_warn_message
ENH: revem tests_metrics.py
ENH: more of this
ENH: final fix: clear all regisitries
FIX: PPCA related warnings
ENH: context manager + decorator
FIX: doc example
ENH: address discussion
ENH: address discussion
rebase
cosmits
FIX
FIX typo 2
address discussion
remove import
Improve doc
py3k syntax
ENH: color scheme + add PCA
COSMIT
address duscssion
address duscssion 2 + expose modern API
ICA vs PCA barrier free + fixed
COSMIT
Improve wording + COSMITS
Improve wording + COSMITS II
manual subplots adjust
Eddy L O Jansson (1):
Wrong format specifier used when formatting exception message.
Eduardo Ariño de la Rubia (1):
Fixed incredibly minor spelling mistake
Eric Jacobsen (3):
ENH: Call plot twice per class label rather than for every point.
ENH: Call plot twice per class label rather than for every point.
Merge branch 'plot_dbscan' of https://github.com/ericjster/scikit-learn into plot_dbscan
Erik Shilts (2):
DOC missing parts of docstrings
DOC Remove target_names from boston dataset object description
Erwin Marsi (3):
Added cosine distance metric for sparse matrices
added missing assert in unit test
Fixed doc string; compute cosine distance without copying matrix.
Ethan White (1):
Fix typo in linear_model documentation
Eustache Diemert (60):
first draft of scaling_strategies.rst - still wip
improved scaling_strategies.rst, inc. working links
added Federico as author
cosmit pep8/257
added link to user guide + small cosmit
fix re: ogrisel comments
added note on learning rate evolution
readded a paragraph that got lost in the rebase
COSMIT pep8/257
COSMIT typos
COSMIT typos
added links to plots in narrative docs
added bestofmedia logo
added bestofmedia testimonial
cosmits
cosmits (2)
COSMIT more typos/precisions
inital checkin
first plot working
working boxplot
pep8/257 fixes
more pep8/257 fixes
1st draft of merformance docs
Merge remote-tracking branch 'upstream/master' into pred_latency
some substance to the perf docs
added influence of n_features; added throughput explanation/graph
Merge remote-tracking branch 'upstream/master' into pred_latency
wip sparsify()
Merge remote-tracking branch 'upstream/master' into pred_latency
completed sparsity sections; added benchmark
cosmit link
cosmit double backticks
precisions from @ogrisel
Merge remote-tracking branch 'upstream/master' into pred_latency
Merge remote-tracking branch 'upstream/master' into pred_latency
@ogrisel precision on sparse + elasticnet tradeoff
link to benchmark in github
fix for np.count_nonzero not present in Numpy < 1.6
whats_new.rst
s/speed/latency/g
added section on model complexity
now compares linear vs svr vs rf
added example and fullfilled section on influence of model complexity
removed CV code to tune hyperparameters (better for CI)
mentionned training throughput
added ElasticNet to the narrative doc
more descriptive legend for ElasticNet
merge doc/whats_new.rst
Merge remote-tracking branch 'upstream/master' into pred_latency
found a good elasticnet example
fixed link & desc in WN.rst
added model complexity in latency/throuhputs example plots
added prediction time plot to Reuters example
typos in narrative doc
Merge remote-tracking branch 'upstream/master' into pred_latency
Joel's feedbacks
Merge remote-tracking branch 'upstream/master' into pred_latency
Merge remote-tracking branch 'upstream/master' into pred_latency
added model re-shaping + compressed verbose §
emph. reshaping benefits on I/O
Fabian Pedregosa (1):
Add artwork with reference to logos in about.rst
Federico Vaggi (2):
ENH: Added comparison of other classifiers using partial fit methods
Fixed misc style changes suggested by Ogrisel
François Boulogne (1):
DOC: minor improvement in comments of an example
Félix-Antoine Fortin (2):
DOC fix roc_curve docstring.
DOC: fix 2 covariance examples rst math markup.
Gael Varoquaux (217):
Merge pull request #7 from larsmans/sklearn
MISC: update whats_new
MISC: fix reference to example
DOC: DBSCAN misc doc formatting
DOC: also point installation menu to stable
MAINT: remove sklearn.test()
MISC: deprecation notice
DOC: reduce the number of examples
MISC: document sklearn.test deprecation
ENH: custom distutils clean command
DOC: layout tweaks
DOC: bigger menu fonts
DOC: button layout tweak
TST: avoid a crash in Windows + Anaconda Py3.3
MISC: fix wrong timing in example
TST: avoid nose running sklearn.test as a test
MAINT: randn on float is deprecated
MISC: deprection is in 2 releases
DOC: fix CSS bug
MAINT Update mailmap
REL: update whats_new and version
MAINT/DOC: bump docs and rev numbers to 0.15-git
DOC: link to documentation, not main page
TST: fix test on scipy dev version
MISC: switch line returns back to unix
BUG: restore setup.py clean functionality
MISC: fix rst
COSMIT: fix docstrings in affinity_propagation
ENH: add single_linkage
ENH: single linkage with Cython
ENH: speed up single linkage
ENH: start of complete linkage
ENH: first sketch of weighted linkage
BUG: fix bug introduced in Ward
ENH: add average linkage
BUG: remove debug print
WIP: MST for single linkage
Cosmit
MISC: refactor Ward and linkage in same object
ENH: hierarchical clustering example
DOC: nice hierarchical clustering example
API: rename to agglomerative clustering
Add cython code
MISC: manual merge of master
MISC: Take in account @NelleV's comments
TEST: increase test coverage
MISC: move fast_dict to utils
MISC: new formulation in message
COSMIT: remove unused import
ENH: add single_linkage
ENH: single linkage with Cython
ENH: speed up single linkage
ENH: start of complete linkage
ENH: first sketch of weighted linkage
BUG: fix bug introduced in Ward
ENH: add average linkage
BUG: remove debug print
WIP: MST for single linkage
Cosmit
MISC: refactor Ward and linkage in same object
ENH: hierarchical clustering example
DOC: nice hierarchical clustering example
API: rename to agglomerative clustering
Add cython code
MISC: manual merge of master
MISC: Take in account @NelleV's comments
TEST: increase test coverage
MISC: move fast_dict to utils
MISC: new formulation in message
Merge branch 'hc_linkage' of github.com:GaelVaroquaux/scikit-learn into hc_linkage
ENH: better error message
ENH: add the cosine distance to paired distances
ENH: different metrics in hierarchical cluster
BUG: Remove UTF8 character
MAINT: remove utf-8 headers
WIP
Merge pull request #2418 from rsivapr/typo-docs
Merge pull request #2404 from agramfort/fix_probabilitic_pca
DOC: add Rangespan testimonial
DOC: add new testimonial
MAINT: Add COPYING and README to MANIFEST
Merge pull request #2424 from sergiopasra/no-shebang
DOC: fix bestofmedia testimonial
Cleanup
ENH: address @larsman comments on fast_dict.pyx
MISC: improve tests of fast_dict
Merge pull request #2462 from kmike/metrics-unicode
BUG: fix typing
BUG: fix types
Clean: Separate hierarchical code from fast_dict
Merge pull request #2527 from shoyer/permutation_test_score-docstring
Merge pull request #2505 from rphlypo/master
Merge pull request #2525 from rmcgibbo/hmmfix2
Merge pull request #2517 from jaquesgrobler/simpler-collapsible-toctrees
Merge pull request #2533 from samuelstjean/patch-1
Merge pull request #2543 from johncollins/dev
Merge pull request #2565 from change/change_org_testimonial
MAINT: make gen_rst more robust (minor)
BUG: ward does not take an affinity
DOC: example metric + hierarchical clustering
WIP example l1 metrics hierarchical cluster
BUG: fix partial_fit of MiniBatchDictLearning
DOC: update whats_new
Merge pull request #2615 from rgommers/fix-numfocus-link
Merge pull request #2620 from jaquesgrobler/master
BUG: fix convergence check in OMP
Merge pull request #2488 from oddskool/pred_latency
Merge pull request #2626 from jnothman/string_objects
Merge pull request #2653 from jaquesgrobler/remove_ellipticenvelope_deprecation
Merge pull request #2654 from arjoly/metrics-0.15
Merge pull request #2667 from kowalski87/master
DOC: add phimeca testimonial
Merge pull request #2671 from agramfort/fix_f_oneway_int
BUG: OneVsOneClassifier was broken with string labels
COSMIT: simplify code
Merge pull request #2674 from GaelVaroquaux/bug_ovo_string_y
Merge pull request #2702 from glouppe/imputer-copy
Merge pull request #2714 from amormachine/master
Typo in examples
COSMIT: remove unused import
Merge pull request #2760 from AlexanderFabisch/check_error_patterns
TST: add Py2.6 on travis
BUG: assert_raises_regexp for Py2.6 compatibility
TST: get travis to find the system packages
TST: try to get travis to work with Py2.6
TST: get travis working on Py26 and Py27
TST: Add Python 3.3 to travis
Merge pull request #2767 from GaelVaroquaux/my_master
Merge pull request #2779 from jnothman/ref_sec_link
DOC: some underlines were too short
Merge pull request #2781 from blagarde/patch-1
Merge pull request #2782 from blagarde/patch-2
Merge pull request #2776 from DanielWeitzenfeld/master
Merge pull request #2775 from stefan-w/python-3-fix
WIP: fix StratifiedShuffleSplit
BUG: StratifiedShuffleSplit not obeying n_train
BUG: avoid same indices in test and train
COSMIT
Merge pull request #2791 from adrinjalali/master
Merge pull request #2794 from cli248/master
Merge pull request #2799 from charlescearl/charles-label-prop-doc-updates
Merge pull request #2797 from cli248/fix-DeprecationWarning
DOC: more context in embedded code
DOC: more robust compilation and CSS
Merge pull request #2768 from robertlayton/meanshiftdoc
Merge pull request #2820 from ankit-maverick/issue2819
MISC plot_rfe_with_cross_validation: better comments
Merge branch 'master' into hc_linkage
BUG: different distance names in scipy
BUG: fix minor floating-point precision detail
ENH: Py3 support
BUG: fix tests under Python 3
DOC: improve the clustering metric example
DOC: AgglomerationClustering doc and examples
BUG: _alpha_grid undefined symbol
DOC: Agglomeration clustering docs and examples
DOC: connectivity in agglomerative clustering
DOC: finish touches to hierarchical docs
DOC: fix rest syntax
MISC: typos in docs
MISC: remove reference to deprecated Ward
MISC: replace 'assert' with actual ValueError
MISC: increase test coverage and cosmetics
Merge pull request #2830 from eltermann/wrt-abbreviation
Merge pull request #2750 from jnothman/no-str
MISC: better error messages in BaseLibLinear
MISC+DOC: documentation for FeatureAgglomeration
Merge pull request #2859 from eltermann/fixed-typo
Merge pull request #2864 from sciunto/doc
COSMIT: many cosmetic comments
FIX: fix failing doctest
DOC: link FeatureAgglomeration better to dim red
MISC: documentation and cosmits
BUG: work aroundd Cython bug to build with clang
Merge pull request #2865 from hanke/boston_13features
Merge branch 'master' into hc_linkage
Merge pull request #2877 from pmandera/tfidf-citation
Merge pull request #2917 from ariddell/patch-1
TST: test utils.extmath more robust
MAINT: remove our solve_triangular
MAINT: commit generated code
MAINT: scipy 0.10 and 0.11
DOC: rmk different feature scaling in agglomeration
Merge pull request #2199 from GaelVaroquaux/hc_linkage
Merge pull request #2929 from ajtulloch/izip-tests
BUG: avoid NaNs in arrays passed to scipy.linalg
DOC: docstring formating
Merge pull request #2943 from jyu-rmn/patch-1
Merge pull request #2941 from perimosocordiae/patch-1
Merge pull request #2938 from jnothman/clean_impute
Cosmit
MAINT: point out HMMLearn
Merge pull request #2955 from larsmans/less-blas
DOC: add OkCupid testimonial
Merge pull request #3003 from jnothman/hyperlinked_tt_colour
BUG FeatureAgglomeration meaningless for no samples
TST: more robust test of dropping in Lars
TST: relax warning class checked in lars
TST: lars tests: more brutal ill-conditionning
Merge pull request #3074 from msalahi/iris-file-warnings
Merge pull request #3066 from denisgarci/agglomeration-transform-constructor
Merge pull request #3096 from ash211/patch-1
Merge pull request #3101 from fabianp/logos
BUG: test_lasso_lars_vs_lasso_cd_ill_conditioned
COSMIT: rng.permutation rather than rng.shuffle
Merge pull request #3104 from MechCoder/swap_sparse
BUG: fix doctests
DOC: fix restructured text
Remove HMM examples
Revert "DOC : really fix alpha docstring in dict learning"
COSMIT: remove unused import
MISC: use precomputed variable
MISC: update comment in source
Merge pull request #3193 from mjbommar/issue-3167-eradicate-todense
Merge pull request #3157 from mjbommar/isotonic-increasing-auto
Merge pull request #3199 from mjbommar/isotonic-out-of-bounds-2
Optim: small optimizations
Gilles Louppe (119):
What's new: put major 0.14 additions in first
FIX: remove dupplicate entry in What's new
Polishing on "Who's using scikit-learn"
Website: bottom buttons
Merge pull request #2361 from rolisz/patch-1
FIX: release reference to X in _tree.splitter
Merge pull request #2363 from glouppe/tree-memory
Merge pull request #2385 from ndawe/tree
DOC: docstring for predict_log_proba was wrong
Typo
Merge pull request #2394 from ndawe/export_graphviz
FIX: free self.estimators_
Merge pull request #2430 from glouppe/forest-2414
Merge pull request #2432 from jwkvam/docs_cluster
ENH bagging meta-estimator
ENH: move _partition_estimators to ensemble.base
TST: test_base in sklearn.ensemble
BUG: do not force base estimators to inherit from sklearn in AdaBoost
TST: fix random_state in test_weight_boosting
DOC: base_estimator_ + estimators_ in ensemble.base
COSMIT: tidy up forest code (inspired by bagging)
What's new: bagging
FIX: logaddexp(-inf, -inf) == -inf and not NaN
Merge branch 'master' of github.com:scikit-learn/scikit-learn into logaddexp-fix
DOC: add docstring to _logaddexp
DOC: bootstrap=True by default
ENH: added PresortBestSplitter
ENH: remove assert
COMIST: shorten some long lines
DOC: added entry to What's New
Merge pull request #2469 from glouppe/tree-presort
FIX: don't bias random feature selection
FIX: random number generator
FIX: module RAND_R_MAX
TEST: improve test_distribution
TEST: improved test_distribution
FIX: use XorShift random number generator
DOC: remove rand_r license
TEST: add tests for splitter="presort-best"
Merge pull request #2474 from glouppe/tree-tests
FIX: unsigned int -> UINT32_t
ENH: small optimization in PresortBestSplitter
Merge pull request #2500 from ndawe/r2
ENH: remove offset calculations
ENH: use memset and memcpy when possible
ENH: make sort an inline nogil function
ENH: make node_reset/split/value gil-free
ENH: use npy_int32 type
Merge pull request #2546 from StevenMaude/patch-1
Merge pull request #2550 from StevenMaude/minor-typo-fixes
Merge pull request #2566 from Manoj-Kumar-S/constant-output
What's new: DummyClassifier now supports constant output
COSMIT + FIX in _utils.pyx
Merge commit 'fbab2cd' into gbm
COSMITS + FIX to impurity_improvement
FIX: impurity_improvement
TEST: disable test_32bit_equality
Update max_depth in DepthFirstSplitter + Release splitter in BestFirstSplitter
Add max_leaf_nodes to forest + DOC
Cleanup criteria
Nitpicks
Normalize variance
Merge pull request #2684 from blagarde/master
Merge pull request #2685 from bicycle1885/fix-oob-score
ENH Release gil in _tree.pyx:predict
ENH Use backend="threading" for forest regression as well
FIX: Ensure expected behaviour when copy=True|False in Imputer
FIX: Don't copy X at imputation in the sparse case
COSMIT: remove useless import
Update docstring + Use np.diff
Merge pull request #2710 from eshilts/DOC-load_boston
FIX: don't force finite in imputation
FIX: use _assert_all_finite
Merge pull request #2712 from glouppe/imputer-copy
Pre-initialize all trees before dispatching
Merge pull request #2724 from glouppe/forest-2721
Regenerate C files
Merge pull request #2745 from arjoly/mse-criterion
COSMIT: PEP8 + coding style
COSMIT: PEP8 in test_gradient_boosting.py
COSMIT: Fix some PEP8 in gradient_boosting
COSMIT: PEP8 in test_imputation
COSMIT: Various PEP8
Merge pull request #2790 from jnothman/tree_without_cycles
ENH Make PresortBestSplitter cache friendly + cosmetics
Merge pull request #2800 from eltermann/doc-fix-tree
FIX: Weight impurity decreases with weighted_n_samples
TEST: add non-regression test
ENH: add weighted_n_node_samples property
Cleanup
FIX: Recompile _gradient_boosting.c
TEST: check with scaled sample_weight
Merge pull request #2836 from glouppe/tree-importances
DOC: Entry for #2835 + Put together entries wrt. trees and ensemble
What's new: group entries by topic
ENH: small optimization of the classification criterion
ENH: small optimization to the regression criterion
DOC: update examples, use matplotlib.pyplot instead of pylab
ENH: add explanatory coments
Merge pull request #2964 from abhishekkrthakur/patch-1
TEST: fix test_forest:test_boston (#2965)
DOC: the search for a split does not until one valid partition is found
Merge pull request #2966 from maheshakya/feature_selector
DOC: wordings
FIX: Set correct impurity values in BestFirstBuilder
PEP8
Merge pull request #3053 from ajschumacher/patch-1
FIX: weight impurity improvement by the size of the node
DOC: better doc for impurity_improvement
ENH: declare weighted_n_samples
ENH: rename Node attributes to avoid confusion
ENH: rename normalizer -> weighted_n_samples
ENH: renamed things too fast...
ENH: better API for passing weighted_n_samples
ENH: remove unused variables
Merge pull request #3056 from glouppe/tree-bestfirst
Merge pull request #3078 from ndawe/examples
Merge pull request #3076 from msalahi/sparse-bagging-tests
Merge pull request #3156 from IvicaJovic/fixes
Guilherme Trein (1):
Fix exception message when cloning a estimator that does not implement 'get_params' method.
Hamzeh Alsalhi (19):
Included sprse testing functions test_sparse_classification() and test_sparse_regression() in test_weighted_boosting.py, based on the corresponding testing functions found in test_bagging.py
Removed densification calls on X '.asarry()' from all predict methods, removed 'dense' parameter from check arrays call, removed contiguous cast of the input array X, detected X as list or array to find the number of examples accordinly using len() or shape[0]
In functions def test_sparse_classification() and test_sparse_regression() added tests for coo, lil, and dok sparse matricies. Removed parameter set feautre since there are not parameters that alter sparse/dense data usage. Revised data set to use dataset.make_{regression/classification} for improved test time.
Revised recuring import statement to respect conventions
Updated documention to reflect sparse support on input data
Removed prints statements used for debugging
Revied test_sparse_{regression,classification} to use make_multilabel_classification, revised import statements to adhere to one name per import conventions
Enforced dtype for X, and clarified documentation by indcating the sparse formats supported and how they are treated
Re inserted the check for X.ndim == 2 in initialization, updated a doc detail for X: coo is converted to csr only in the initialization not in the predict functions
Inserted base classifier check for BaseDesicionTree and and BaseForest before enforcing DTYPE in wieghted boosting fit
Resolved python3 forest import error in weighted_boosting.py
Updated documenation for weighted_boosting fit() to indicate behavior of forced dtype
Included testing of the all of the predict/score functions of AdaBoost
Assert sample weight negative sum value error, add authors entry
Fix score function parameter mistake in test_weighted_boosting
Remove exectuable permision bit
Fixed typo toCSR -> tocsr
Trivial doc alteration, travis timeout
Fixed paramaters to make_multilabel_classification and make_regression to keep tests under 1 second, reformated some blank lines
Hendrik Heuer (1):
PY3 fixed examples
IvicaJovic (1):
fix wrong confidence interval
Jacques Kvam (3):
DOC: Update pyamg site and fix a typo.
Remove superfluous AdaBoostRegressor call.
PY3: fix map usage in ensemble.partial_dependence
Jake VanderPlas (21):
DOC: fix metrics documentation format
Merge pull request #2489 from samuela/patch-1
ENH: add PolynomialFeatures preprocessor
address PR2585 comments
import pylab -> import matplotlib.pyplot
TST: fix PolynomialFeatures test error
TST: fix doctests & pep8
DOC: adjust notes in PolynomialFeatures
reorder polynomial features
fix ordering & docs of PolynomialFeatures
TST: remove stray print statement
DOC: typo in kernel formulas
Merge pull request #2608 from jakevdp/master
Merge pull request #2585 from jakevdp/polynomial
DOC: PolynomialFeatures narrative doc
DOC: fix auto-links
DOC: fix Polynomial Regression docs
DOC: fix test failure in linear model
SPD: reuse variable in GaussianNB for speedup
DOC: typo in binary_tree docs
Merge pull request #2765 from AlexanderFabisch/validation_curves
James Yu (2):
DOC: fixed docstring formatting
typo
Jaques Grobler (95):
initial commit
remove placehoder
Merge branch 'tutorial' into DOC_olivier_tut
remove duplicates for merging larsmans branch
initial tutorial re-added with it`s commit history
Placeholder to tutorials page
moved tutorial out of tutorial folder
tutorial index initial touchups
Tutorial Setup
fix and squash broken branch, remove all but section 3 and redo some docstring breaks
fix broken directory name
Merge pull request #2329 from NelleV/website
DOC: few small doc fixes to layout bugs on new website
comments added to the changes
Merge pull request #2331 from jaquesgrobler/master
fix sidebar button size staying constant and leaving a mountain of white-space due to expanding/collapsing toctree
fix cut-off ML map
Merge pull request #2336 from jaquesgrobler/DOC_fix_whitespace_on_userguide
commit added for hack-warning
Merge pull request #2337 from jaquesgrobler/DOC_fix_cut_ML_flowchart
Merge pull request #2345 from rgbkrk/patch-1
first carousel version added
firefox fix and more images added, auto-cycling disabled
arrows switched for dots
have images link to relevant examples
slight layout adjust
small layout changes for firefox, images taken from generated images now
indentation fixes
add more examples and cropping to first image
disable carousel for small displays, small tweaks
merge conflict fix
redo old deletions
contributors logos added to index screen footer
changed hyperlink to funding anchor - added same for the images
old footer removed for index page
removed little underscores appearing between images - caused by link added to images
add about us to community block
Merge pull request #2350 from jaquesgrobler/DOC_inria_google_logos
fix broken logo links on homepage and remove unused duplicate images
another speedup attempt for index.html: carousel images to use thumbnails instead, 1 blocking js removed for page (sidebar.js)
cleanup changes
remove paren
add lossless image compression for docbuilding machine
switch species_kde to thumbnail
logo lossless compressions
add coverage test for travis
coverall added
remove whitespace, fix pip install
adds minutes + seconds to examples
cosmetic - shorten long line
add copybutton to code examples
remove writing of minutes/seconds
remove duplicate line
cleaner toctree collapsing
remove unused images
mention python.org in js file
Merge pull request #2515 from jaquesgrobler/add_minutes_examples
fixes mis-alignment of Documentation link on front page
Merge branch 'master' of github.com:scikit-learn/scikit-learn
SGD and SVC duplicate example sorted
sort out plot_iris vs plot_svm_iris
some cosmetic commits and delete duplicate files
Merge branch 'master' into DOC_olivier_tut
Merge branch 'master' into DOC_olivier_tut
some cleanup, conflict fixes, squashes, re-dos
redo typo fixes
small css tweak
speedup for docbuild using joblib
change cache directory
typo fix
changes to headings and sections
remove face-recognition exercises, skeletons, etc and merge artifacts
further changes to headings and subsections
re-order some Documentation categories in documentation.html
fix broken link for stable version from dev version
fix tutorials index, update dropdown menu in navbar, update stat_learning tutorial link on documentation.html page
typo fix
Merge branch 'master' into DOC_olivier_tut
remove extra headings from tuts page
update what`s new
remove EllipticEnvelop deprecation
Merge pull request #2651 from amueller/spotify_testimonial
whatsnew conflict fix
Merge branch 'master' into DOC_olivier_tut
recommit joel comments
where to from here section
whitespace removals
olivier suggestions
changed PCA to truncated SVD
MAINT deprecate HMMs
conflict fix
small fixes
Merge pull request #1971 from jaquesgrobler/DOC_olivier_tut
sphinx colon error fix
COSMIT fix PEP8 errors
Jean Michel Rouly (1):
Using the 'is' token compares identity; this should not be used for string value comparison.
Jeffrey Blackburne (1):
ENH shuffle option for StratifiedKFold
Jelle Zijlstra (1):
fix defaultdict call
Joel Nothman (97):
DOC fix comment on svm probability param
DOC add missing Returns description
MAINT deprecate indices=False for cross validation generators
COSMIT rewrite precision_recall_fscore_support
FIX warning interaction problem, DOC clarify parameters
FIX ignore corrcoef warning
Use label_binarize instead of LabelBinarizer
Use specialised warning class
TST messages for UndefinedMetricWarning from prf
TST use ignore_warnings helper
TST classification_report example that does not trigger warning
FIX use fixes.bincount for numpy 1.3.0 support
FIX Py2.6-compatible use of str.format
FIX finish deprecation of indices in CV
FIX limit warnings for recall_score, precision_score, f1_score,
Merge branch 'master' of https://github.com/scikit-learn/scikit-learn into prf_derivative_warnings
COSMIT use assert_no_warnings where appropriate
FIX stop forcing deprecation warnings for external packages
FIX use of str.format for Py2.6
ENH support multiclass targets of string objects
TST Add benchmarking script for multilabel metrics
DOC Add what's new entry for #2626
COSMIT in response to feedback
Merge pull request #2660 from amueller/mbkmeans_compute_labels_fix_new
FIX suspected bug for non-hypercube make_classification
FIX Low-memory cluster sampling in make_classification
FIX don't leave clusters in uint8
Remove debug print statement
FIX update data-dependent test assertion
Use internal sample_with_replacement rather than Python's random.sample
TST/COSMIT cleaner make_classification and fix tests
DOC add missing comment for C parameter
DOC remove duplicate reference to example
FIX Use Py2.6-compatible string formatting
DOC add learning_curve to what's new and API reference
FIX explicit type for offset variable
FIX remove confusing BaseEstimator.__str__
ENH/FIX Change Tree underlying data structure
DOC note #2732 in what's new
DOC fix documentation to match default value
ENH remove unnecessary CSR->CSC transform in text feature extractors
Merge pull request #2772 from jnothman/tfidf_no_copy
DOC make indifferent to compressed sparse orientation
DOC add link to module reference from class/function pages
FIX remove reference cycles from Tree
FIX dtype refcount; block resizing after building
Remove 'locked', add comments
COSMIT remove unused imports and variables
Merge pull request #2818 from jnothman/flakes
Merge pull request #2826 from perimosocordiae/patch-1
Merge pull request #2840 from earino/develop
Merge pull request #2869 from jperla/master
STY Simplify and avoided numpy warnings in imputation
Merge pull request #2925 from ajtulloch/izip-cross-validation
ENH/TST remove unnecessary sorts and complete testing for sparse median
COSMIT Remove unnecessary assertion
FIX for numpy compatibility, code clarity
MAINT warn of future behaviour change proposed in #2610
FIX for early numpy where astype(..., copy=...) unavailable
DOC Remove redundant parameter docs
Merge pull request #2991 from Manoj-Kumar-S/csr_sparse_center
DOC use hyperlink colour when in tt
Merge pull request #3010 from griffinmyers/fix-select-k-best
TST/COSMIT tests/comments for [sparse]center_data
Merge pull request #2952 from jnothman/future_warn_two_labels
Merge pull request #3057 from ajschumacher/patch-4
DOC Comment on thresholds[0] for roc_curve
DOC a note on shuffling for cross-validation
DOC shuffled StratifiedKFold now in whats_new
Merge pull request #3133 from kmike/patch-2
Merge pull request #3134 from jrouly/string_values
Merge pull request #3140 from MechCoder/small_fixes
Merge pull request #3192 from mmaker/master
DOC fix typo
Merge pull request #3196 from andrewclegg/master
Merge pull request #3197 from mjbommar/issue-3167-eradicate-todense
Merge pull request #3198 from apw/patch-1
Merge pull request #3208 from staubda/improve_RFE_doc
Merge pull request #3217 from bwignall/innerprod
Merge pull request #3225 from bwignall/quickfix-typo
MAINT deprecate sequences of sequences support
FIX remove duplicates in MultiLabelBinarizer
COSMIT use MultiLabelBinarizer in LabelBinarizer during deprecation
COSMIT use mlb for MultiLabelBinarizer instances
ENH do not allocate memory for temporary array of 1s
DOC remove mention of sequence of sequences in Parameters sections
FIX don't use dict comprehension for Py 2.6
FIX No set construction shorthand in Py2.6
DOC comment on _transform interface
TST Validate input MultiLabelBinarizer.inverse_transform
No set construction shorthand in Py2.6
DOC/FIX Address @arjoly's comments
TST stronger test for non-integers in MultiLabelBinarizer
Assert or ignore all sequence of sequences deprecation warnings
TST avoid more warnings related to sequence of sequences
TST fix testing for sequence of sequences warning in metrics
Merge pull request #3240 from JelleZijlstra/fix-defaultdict
Johannes Schönberger (134):
Add implementation of RANSAC algorithm
Add ransac function to __init__.py of utils package
Fix bug in residual determination
Add step-by-step description of RANSAC algorithm
Add parameter description to RANSAC and modify API
Add example plot script for RANSAC
Add simple RANSAC unit test
Add unit tests for RANSAC is_model_valid and is_data_valid
Add unit tests for RANSAC max_trials option
Return number of trials in RANSAC
Fix n_trials return value and add additional test
Add n_trials return value to doc string
Add test for stop_n_inliers parameter
Add test for stop_score parameter
Move ransac.py to _ransac.py to avoid nosetest namespace conflict
Set numpy random seed for all test functions
Remove unused variable
Add missing information predict method of estimator
Skip iteration for empty inlier sample set
Move algorithm description to notes section
Add some reference papers
Fix estimator parameter description
Add empty line between parameter and return value description
Move RANSAC implementation to linea_model subpackage
Implement RANSAC as estimator class
Fix indentation
Add score method
Add RANSAC to __init__ of linear_model sub-package
Add author and license info in source file header
Define source file encoding
Fix deprecated naming of estimator object
Update unit tests with new estimator class interface
Update RANSAC example script for new estimator class interface
Add doc string for predict method
Add doc string for score method
Set default initialization parameters
Fix default parameters of RANSAC estimator
Add missing trailing underscore for estimator attributes
Remove deprecated ransac implementation
Add description for min_n_samples parameter
Move estimator_, n_trials_, inlier_mask_ initialization to fit method and use
Add extra line between doc string sections
Add support for absolute and relative min_n_samples
Add default values for min_n_samples, residual_threshold
Include random_state utility
Use random_state in test cases
Explain behavior when no base estimator is specified
Fix test case function name
Add specific test case for score method
Add specific test for predict method
Mark all parameters as optional
Add missing reference papers
Set random state in fit rather than __init__
Add specific exception description for ValueError
Change indentation of multiline if-statement
Remove unused variable
Use term callable instead of function in doc string
Add more precise description of is_*_valid parameters
Combine nested if-statements
Remove default perceptron base estimator for integer data type
Add support for sparse feature vectors
Fix indentation of multi-line if-statement
Remove trailing empty line after doc strings
Add test case for sparse feature matrix
Fix flake warnings
Add test case for case without specified estimator
Add test case for min_n_samples parameter
Fix description of default base estimator behavior and add corresponding test case
Add description of RANSAC to linear-model docs
Move ransac example to linear-model folder
Add meta-data title, author and year to references
Fix indentation
Change ransac docs section title
Remove academic detail description
Remove duplicate detailed description from doc string
Change default parameters, so example works without specific parameters
Add support for multi-dimensional target-values
Add reference to narrative documentation for detailed description of RANSAC
Remove empty line
Add more descriptive explanation for raised ValueError
Fix bug in is_data_valid test
Add more specific test for is_data_valid and is_model_valid functions
Stylistic multi-line statement change
Improve ValueError description
Remove email adress
Fix typo
Set random state of base estimator as well
Use MAD as default residual threshold
Use np.logical_not rather than tilde
Extend RANSAC example with comparison to BaggingRegressor
PEP8 indentation fix
Remove double-space
Add more detailed description for score method
Improve description of base_estimator parameter
Add tests for more types of sparse matrices
Derive Ransac also from MetaEstimatorMixin
Document use cases of is_*_valid functions
Improve and document n_trials_ attribute
Fix parameter description of X and y
Change naming scheme of variables for consistency
Fix return value description of predict
Fix y paramter description of score
Change linear_model access and call explicitly with kwargs
Remove bagged regressor
Add missing parameter of score method
Unwrap lines
Make sure min_n_samples is not larger than number of samples
Remove _n_ from min_samples
Sample subset without replacement
Add reference to max_trials
Rename min_samples
Improve explanation of degenerate case
Fix typo
Improve RANSAC example script
Make sure dimension is correct for all estimators
Add note about computational cost of is_*_valid functions
Rename RANSAC to RANSACRegressor
Rename base_estimator to estimator
Use LinearRegressor as default estimator for all data types
Add RegressorMixin base class
Raise ValueError for non-integer absolute min_samples value
Rename base_estimator to estimator in docs
Clarify default estimator by explicitly instantiating the class
Remove unused outliers parameter
Rename estimator to base_estimator
Add additional floating point test for absolute min_samples value
Use assert_equal, assert_less rather than plain assert statement
Add RANSACRegressor to whats-new doc section
Add test for default value of min_samples
Add test for invalid value of min_samples
Add test for custom residual_metric
Add test for default residual threshold
Remove reference to web page
ENH: Add dynamic maximum trial determination to RANSACRegressor
John (1):
adding conditional import to lfw.py for python3 support
John Schmidt (1):
DOC reference for k-means++ in clustering narrative
Jorge Cañardo Alastuey (1):
BUG: Compare strings for equality, not identity.
Joseph Perla (1):
Fix typo in fast sgd classifier implementation comments
Joshua Vredevoogd (1):
DBSCAN BallTree implementation
José Ricardo (1):
Fixing small typos in the docs.
Julien Miotte (1):
Without pl.show, the figure won't be displayed.
Karol Pysniak (3):
Added unit tests for parameters in isotonic regression
Fixed style issues as indicated by pep8
Fixed passing y_max and y_min to IsotonicRegression in test_isotonic
Kemal Eren (1):
skip perfect checkerboard test
Kenta Sato (1):
Fix OOB score calculation for non-contiguous targets
Kyle Kastner (12):
Updated with more explanatory text. Also changed to use train_test_split function
Corrected some typos, and added more explanation for the precision recall example.
Clarified a few wordy sentences
Attempted a clearer explanation of precision-recall score
Adjusted to fit 80 character columns, and corrected definition of precision
Removed random_seed argument to svm.SVC
Added contributions to whats_new
Further expansion of the precision_recall curve explanation, spelling corrections, and general cleanup.
Added random_state argument to svm.SVC
Added links to cross_val_score, auc_score, and recall_score
Added link to precision_score
Updated whats_new and fixed a typo in plot_precision_recall.py
Kyle Kelley (1):
Converted Markdown style link to restructured text
LK (2):
Update dpgmm.py
Update dp-derivation.rst
Lars Buitinck (258):
There is no scikits.learn.feature_extraction.text.sparse
Fix minor typo
Spelling error
Add a small part about cross-validation + copyedit
Copyedit on the "working with text" chapter
Text chapter: load_files renamed load_filenames
Merge branch 'master' of github.com:scikit-learn/scikit-learn-tutorial
Revised text classification chapter
Copyright and project name in HTML footer
Updated README
Remove ugly BLANKLINE stuff from text chapter
Use grid_scores_ instead of _get_params + small fixes in text chapter
scikits.learn -> scikit-learn + url of Numpy
s/scikits\.learn/sklearn/g
update cluster docs (DBSCAN)
bring text classification somewhat closer to current API
DOC move old tutorial out of the way for merge
Merge branch 'tutorial'
DOC credit to @ephes and myself for log loss in metrics
DOC fix example comments
DOC typos involving Nyström
DOC copyedit SGDClassifier docstring
DOC improve docstrings in sklearn.base
FIX integer types in Ward clustering
MAINT deprecated ENet param used in doc and benchmark
MAINT remove deprecated parameters (the easy cases)
FIX rm spectral_embedding import from sklearn.cluster
DOC typo
FIX + COSMIT Reuters out-of-core example
FIX py2.6 compat in biclustering example
MAINT remove Counter from fixes; no longer used
COSMIT refactor Hungarian algorithm
DOC while we're at it, link to our own RBM docs
DOC make DictVectorizer docstring refer to FeatureHasher
FIX remove warnings from univariate FS
ENH add VarianceThreshold feature selection method
Merge pull request #2308 from pprett/gbrt-check-supported-loss
DOC clarify distances in KMeans' _labels_inertia
COSMIT skip some repeated computations in k-means
DOC ASCII only in docstrings
ENH speed up NMF (about 30% off topic extraction runtime)
ENH export randomized_svd publicly
Merge pull request #2210 from pgervais/distances_argmin
Revert "CSR matrix support in pairwise_distances_argmin_min"
wip
ENH prettier output from NMF example
FIX failing test in NMF due to negative zeros
DOC broken link in HMM narrative
ENH micro-optimize pairwise_distances_argmin_min
BUG duplicate finity check in input validation
MAINT dedup Brandyn A. White in mailmap
MAINT remove useless deprecation in sklearn.utils
BUG rename n_iterations to n_iter in TruncatedSVD
ENH SGD Cython improvements
DOC boolean masks in CV generators are deprecated
DOC what's new: fast_dot is internal, so don't mention it
Merge branch 'bagging'
DOC copyedit DBSCAN implementation notes
DOC output types should never be "array_like"
DOC+COSMIT: typos, lots of them
ENH: don't call astype when a copy is not needed
BUG oops, *safe_* asarray
DOC improve encoding docs
MAINT remove deprecated code from CD
Partially revert "MAINT remove deprecated code from CD"
TST speed up biclustering tests
MAINT remove dead code from LibSVM
ENH 10x speedup in dump_svmlight_format
DOC: NMF narrative: describe optimization problem
Merge pull request #2426 from larsmans/sgd-improvements
Merge pull request #2457 from untom/rbm_csr_format
MAINT zap unused import
DOC optimized PNG images with OptiPNG
DOC remove qda and faces from website carousel
DOC replace jquery by minified version
Merge pull request #2479 from larsmans/website-speed
DOC typo: polynominial
MAINT list @paulgb's full name
DOC improve biclustering docstrings
ENH sparse matrix support in pairwise + optimizations
ENH optimize NMF inner loop
COSMIT pyflakes feature_extraction.text tests
DOC error in feature_extraction.text docstrings
MAINT add authors to validation.py and pairwise.py
MAINT: use python setup.py clean in Makefile
DOC docstring for extmath.norm
ENH refactor squared-norms computation to extmath
ENH use row_norms in KDE code
FIX fit followed by partial_fit in multiclass SGD
MAINT use subprocess.call, not os.system
COSMIT remove dead code in k-means
Merge branch 'refactor-squared-norms'
ENH honor Y_norm_squared when X=Y in euclidean_distances
COSMIT use norm function in feature selection
MAINT simplify f_oneway in feature selection
DOC norm optimizations in what's new
MAINT remove mlcomp document classif. example
ENH optimize one more sq. distance in k-means
TST fix broken fast_dot test
COSMIT friendlier output from faster NMF benchmark
ENH micro-optimize NMF inner loop
TST fix still broken test_fast_dot
DOC sparse matrix support in BernoulliRBM
FIX sparse matrix indexing in BernoulliRBM
TST clean up after ourselves in SVMlight test
COSMIT micro-optimize norm computation in NMF
DOC fix cross-decomposition docstrings
DOC remove confusing comment from TruncatedSVD
DOC: multiclass: make the warning more prominent
DOC improve feature selection docs
COSMIT remove useless "if False" in kernel approximations
DOC explain ARPACK algorithm in TruncatedSVD docstring
BUG SVMlight loader should check whether n_features is big enough
BUG lower space complexity of estimate_bandwidth to linear
DOC be more explicit on mean-shift scalability
Merge pull request #2541 from dengemann/fix_dot
MAINT: use $(MAKE) for recursive make
DOC add sklearn.base to generated docs
ENH use fast row_norms helper in preprocessing.normalize
ENH use fast row_norms in dictionary learning
BUG don't densify sparse matrix in BernoulliRBM.score_samples
ENH speed up progress reporting in RBM
ENH better error message when CountVectorizer prunes away all terms
TST disable non-doctests in Comp. Perf. docs
DOC {min,max}_df look at document freq, not term freq
DOC fix BernoulliRBM._fit docstring
Merge pull request #2642 from larsmans/rbm-speedup
ENH: faster cartesian product in make_classification and document complexity
Revert "ENH: faster cartesian product in make_classification and document complexity"
MAINT remove some dead code from the LibSVM wrapper
MAINT use TruncatedSVD in pipeline tests
DOC optimize Phimeca logo, size halved
Merge pull request #2673 from larsmans/libsvm-310
ENH filter out zeros early in FeatureHasher
FIX PCA.score_samples didn't do input validation
FIX MBKMeans w/ explicit centers and n_init>1, part 2
ENH show top term per cluster in doc. k-means example
FIX n_init bug in k-means
ENH make_pipeline and make_union utility functions
MAINT refactor fast_dot
ENH speed up SVMlight loader using Cython's array support
COSMIT get rid of deprecation warning in tree tests
MAINT ignore sklearn/tree/_utils.c in diff
COSMIT tree: unused variable warnings and use for/range
BUG fix unchecked mallocs in trees
Merge pull request #2715 from larsmans/tree-malloc
Merge pull request #2734 from eloj/cv-broken-format-string
MAINT remove last trail of the ArrayBuilder
DOC clarify order of output in NB predict_proba
FIX don't put data in source dir in bench_covtype
Merge pull request #2754 from eltermann/doc-fix-tfidf
DOC + FIX mean_ in PCA
ENH faster heapsort in trees
ENH introsort in tree learner
Merge pull request #2747 from larsmans/tree-sort
DOC logistic regression attribute docs + authorship
DOC up the sales pitch for SGD
COSMIT pep8 + full stop police
FIX error message with sparse precomputed kernels
Revert "FIX error message with sparse precomputed kernels"
FIX error message with sparse precomputed kernels (second try)
MAINT optimize Spotify logo (- a few hundred bytes)
FIX restore _pairwise on SVMs
ENH sparsify and densify methods for CD models
MAINT comment out unused import in example
MAINT full name for @h10r
FIX error message from trees for large inputs
Revert "ENH sparsify and densify methods for CD models"
MAINT missing import_array() in isotonic r. Cython code
MAINT get rid of compiler warnings from Liblinear
DOC no more "arithmetical order" for classes
FIX integer dtype for labels in DBSCAN
FIX numerical stability issue in BernoulliRBM
ENH speed up RBM training with scipy.special.expit
ENH re-instate extmath.logistic_sigmoid
FIX DictVectorizer handling of empty inputs
TST older nosetests compat in DictVectorizer test
Merge pull request #2882 from larsmans/expit
FIX predict_proba status on SGD and SVC when disabled
ENH use hasattr "predict_proba" in bagging
FIX one more unchecked malloc in the tree code
DOC SVC.predict_{,log_}proba does not return X
DOC what we call poly features are called interaction features in stats
FIX OneHotEncoder: check value max when n_values is integral
Merge pull request #2910 from larsmans/hasattr-predict-proba
Merge pull request #2876 from Manoj-Kumar-S/fix_auto
DOC: note about numerical precision in euclidean_distances
TST fix failing doctest for OneHotEncoder
FIX GBRT missing from covertype benchmark usage
MAINT drop support for NumPy < 1.6.1
MAINT drop support for SciPy < 0.9
MAINT remove useless import
ENH more optimizations for RBM
FIX + TST stability problems with scipy.special.expit
FIX input validation in Nystroem
FIX Nystroem input validation, again
FIX error handling in SVM
FIX decouple spectral embedding from TransformerMixin
DOC fix rendering of fetch_mldata example
MAINT remove leftovers from solve_triangular
COSMIT get rid of warning from expit import
COSMIT hinge_loss: better input validation
MAINT remove some deprecated stuff
MAINT final occurrence of "Scaler"
DOC/MAINT: clarify cblas/README.txt
DOC clarify Imputer constructor: arbitrary strings not accepted
MAINT remove deprecated functionality from SGD
ENH use threads instead of multiprocessing in SGD
MAINT use CBLAS instead of Fortran API in Liblinear
BUG restore joblib logging behavior
BUG joblib writes to wrong dir
DOC trees/ensembles: class labels need not be integers
ENH pairwise L1 distances for sparse matrices
FIX TfidfVectorizer to no longer ignore binary param
MAINT don't use Perl in Makefile when sed suffices
DOC heapsort is not stable at all
FIX cross_val_score to take y as a list
FIX cross_val_score to take y as an *optional* list
FIX sed usage in Makefile
FIX Makefile to use Perl again
COSMIT: pep8, trailing spaces
TST don't run fast_dot tests on numpy>=1.7.2 + pep8
DOC: tfidf is actually tf*(idf+1) = tf + tf*idf
DOC: clear up the big elastic net confusion (I hope)
MAINT lazily import scipy.cluster
ENH micro-optimize a few tests
ENH micro-optimize fast MCD
FIX random_state validation on c_step
DOC mention "shape" in AP docs
ENH less copying in validation for neighbors
FIX safe_asarray to handle LIL, DOK formats
DOC: website: feature selection on front page
FIX loss function example
DOC: clustering: merge discussions of k-means and inertia
MAINT remove unsupported documentation formats
DOC: move mean_shift docs to MeanShift
ENH factor out squared norm helper
DOC no input validation in constructors
COSMIT six.{map,range} usage in partial_dependence
DOC random_state is an arg to LogisticRegression
FIX MemoryError raising in trees (+test)
DOC: add PyPy support to FAQ
Merge branch 'simplify-fselection'
MAINT make export_graphviz more exception-safe
Revert "COSMIT skip some repeated computations in k-means"
DOC L1 distance works for sparse matrices
Merge branch 'pr/3120'
FIX numerical stability in GMM with eigh sampling
DOC multiclass: OvO needs predict_proba or decision_function
Merge pull request #3184 from YS-L/tfidfvectorizer_idf
DOC: DBSCAN: there's no calculate_distance function
DOC: improve feature_extraction.text docstrings
TST skip flaky label propagation test
DOC we work with Python, not against it :)
ENH/DOC fix poly features complexity
ENH interaction_only in PolynomialFeatures
Merge pull request #3239 from larsmans/faster-poly-features
FIX PCA error handling for invalid n_components
Laurent Luce (1):
Fix Machine Learning for NeuroImaging in Python link.
Laurent Pierron (1):
Improve the GMM PDF example.
Luis Pedro Coelho (1):
DOC Fix ``copy_X `` default in documentation
Maheshakya Wijewardena (21):
Added reference to function
Added :func: reference
Added :func: reference
Implemented median and constant strategies in DummyRegressor
Depreciated y_mean_ and y_median. Code is formatted according to pep8
Removed extra lines. Added deprecation warnings for y_mean_
Updated document to reflect changes in the DummyRegressor
Fixed value error. Scalar values are allowed for constant when not multioutput
Done refactoring and necessary changes for multiouput constant strategy. Added tests and updated the documentation.
Changed value error messages to be more informative in constant strategy.
Edited the documentation-unordered list
Implemented additional test cases. Removed the check for lists and numpy arrays in constant strategy.
Refactored the code and added quotes
Recreated test cases with random state
Updated tests. Fixed flake8 errors
fixed pep8 errors in dummy.py
Changed to random_state
Changed documentation in DummyRegressor
Removed one of 'is a regressor' part in the DummyRegressor documentation
Added _LearntSelectorMixin in BaseGradientBoosting
Added extra check to feature importances
Manoj Kumar (35):
Constant output dummy classifier
Minor doctest change
Handled exceptions in fit; Added tests
TST Tests for string labels, DOC Minor doc changes
DOC Minor doc changes
Removed unnecessary attributes
ENH: ElasticNetCV and LassoCV raise ValueError with multitarget outputs
Testing log_loss and hinge_loss under THRESHOLDED_METRICS
FIX: Removed redundant code
TST: Test class variance and string input
FIX: Changed str vs float invariance test
ENH: MultiTaskElasticNet (and Lasso) CV
Proper centering of alpha_grid for sparse matrices
FIX: Normalize=True
Fixes Issue 2751
FIX: Removed coef and improved initialization
FIX: sample_weight='auto' for RidgeClassifier
FIX: Label encoding done in compute_class_weight
ENH: Speed up sparse_coordinate descent
Replaced cython calls for dot operations with BLAS calls
Speed up using typed memory views
Preserve CSR storage format when input is CSR in sparse_center_data
ENH: Refactoring and optimisation of sparsfuncs.pyx
ENH: Optimise sparsefuncs
Moved sparsefuncs to sparsefuncs_fast
TST: Added tests for non-CSR/CSR format
ENH: Swap rows in sparsefuncs
Made the following changes
COSMIT: Replaced ptr1/2 with start/stop
Made the following changes
Improved formatting of unsupported sparse matrices in swapping
FIX: ENetCV and LassoCV now accept np.float32 input
FIX: Use coordinate_descent_gram when precompute is True | auto
Remove unused param precompute from MultiTask models
FIX: Raise ValueError for invalid precompute
Mathieu Blondel (62):
BUG: FIX Crammer-Singer formulation in the binary case.
Better test for auc_score.
Optim for Crammer-Singer formulation in binary case.
Completely avoid for loop in _auc.
Typo.
Update my URL.
Better docstring for KMeans.predict.
Move estimate_bandwidth test to its own function.
Add predict method to MeanShift.
Add predict method to AffinityPropagation.
Add missing docstrings to test functions.
Add what's new item.
Remove warning in AffinityPropagation.
Fix grid search test.
More user-friendly error message.
COSMIT: change variable name.
Merge pull request #2368 from emsrc/cosine_distance
Use pairwise_distances_argmin_min function.
Use pairwise_distances_argmin_min in examples.
Add pairwise_distances_argmin.
Add tests for pairwise_distances_argmin.
Merge pull request #2410 from Balu-Varanasi/pep8_fixes
Merge pull request #2411 from Balu-Varanasi/remove_unused_import
Merge pull request #2415 from kemaleren/skip_checkerboard
Add what's new entry for pairwise_distances_argmin_min.
Cosmit: move log_loss.
Merge pull request #2558 from Jorge-C/patch-1
Test correctness of average_precision_score.
Merge branch 'master' of github.com:scikit-learn/scikit-learn
Remove some warnings.
Add docstrings to _auc and _average_precision.
BUG: ValueError was assigned as local variable!
COSMIT: no need for parentheses.
More robust unit tests for fast_dot.
Remove warnings in ProbabilisticPCA tests.
One less warning in QDA tests.
Simplify fast_dot.
Remove print statement.
Add lobpcg in possible solvers.
Fix fragile symmetry check.
Fix fragile test.
Fix typo in test.
Merge pull request #2659 from jakevdp/gaussiannb_speedup
Handle n_features < n_informative case in make_regression.
Merge pull request #2743 from eltermann/doc_fix
Test primal-dual relationship in Ridge.
More robust test for sample_weight in Ridge.
Add mean_absolute_error to scorers.
Merge branch 'master' of https://github.com/scikit-learn/scikit-learn
Fix doctest.
Rename _split_with_kernel to _safe_split.
Rename fit_and_score to _fit_and_score.
Merge pull request #2762 from sergiopasra/fedora-package
Merge pull request #2761 from ogrisel/thanks
BUG: dual_coef must be defined inside the try block.
Merge pull request #2789 from ugurthemaster/patch-2
Merge pull request #2827 from amueller/fix_subgradient_equation
Cosmit in SGD doc.
Merge pull request #2852 from jwkvam/adaboost_test
Add test for _safe_split with pre-computed kernel.
Fix test failure on Python3.
Merge pull request #3055 from ajschumacher/patch-3
Matthew Brett (1):
ENH: use setuptools for bdist_wheel command
Matthias Feurer (1):
ENH: make trees work with very small max_features.
Max Linke (1):
DOC: kmeans runs inits in parallel, not distance computations
Michael Becker (3):
TST: Switch from python 3.3 to 3.4 in travis
TruncatedSVD: Calculate explained variance.
Authors: Update based on #3067
Michael Bommarito (40):
Adding string={pearson, spearman} option to increasing argument in IsotonicRegression.
Adding increasing and decreasing tests for both Pearson and Spearman increasing argument options
Refactoring increasing_bool set into _check_increasing method
PEP8ing isotonic tests
PEPing isotonic regression
Docstring style changes
Change arguments to increasing={'auto', True, False} and default to 'auto'; implement Fisher transform and warning on 0 \in CI
Adding test for CI check and removing Spearman/Pearson-specific tests.
Minor docstring cleanup
Minor docstring cleanup
Docstring fix
Replacing the non-test .todense() methods with .toarray()
Replacing the test .todense() methods with .toarray()
Improving tests to affirm no CI warnings are thrown when appropriate
Matching CI calculation and docstring with reference
Fixing docstring
Refactoring check_increasing, ensuring that increasing_ is only set on fit/fit_transform, and fixing rho \in {-1, +1}
Reorganizing tests to isolate check_increasing from increasing='auto'
Fixing redundant np.asarray(X.toarray()) in metrics
Improving tests based on feedback from @GaelVaroquaux
Improving tests based on feedback from @GaelVaroquaux
Fixing matrix vs. vector notation for X
Fixing space in docstring for default value
Adding check_increasing to classes.rst
Adding no-warning assertions to IR auto tests
Additional .todense() -> .toarray() fixes
Fixing small formatting issue in doctest
Fixing classes.rst
Adding Notes section to img_to_graph and grid_to_graph re: np.matrix->np.ndarray
Adding what's new item for sklearn.feature_extraction.image np.ndarray changes
Switching from np to math for scalar float ops
Merge remote-tracking branch 'upstream/master' into isotonic-increasing-auto
Additional docstring fixes for issue #3167
Merge branch 'master' of https://github.com/scikit-learn/scikit-learn
Adding tests to handle out_of_bounds parameter.
Adding out_of_bounds parameter to handle values outside training domain
Adding test for out_of_bounds argument validation
Reworking .transform conditionals for out_of_bounds argument
PEP8 fix
ENH Dense pipeline support for RandomTreesEmbedding via sparse_output param
Michael Eickenberg (6):
TST safe_asarray for dok_matrix and lil_matrix
Now capable of treating sample_weights in feature space
Ridge regression now can use sample_weights in feature space. Summary commit over around 20 commits to avoid failing tests
updated authors
failing test for wrong solver exception
Added raise statement in ridge_regression solver check
Michael Hanke (1):
BF: load_boston() return 13 features, but 14 feature names
Michele Orrù (1):
Typo.
Mikhail Korobov (10):
TST "confusion_matrix" was a duplicated key in CLASSIFICATION_METRICS and NOT_SYMMETRIC_METRICS dicts
TST style fix: old_error_settings should be outside try-finally block to be safely used in finally statement
FIX classification_report shouldn't fail on unicode labels in Python 2.x
FIX parallel FeatureUnion.fit and fit_transform
TST skip `test_k_means_plus_plus_init_2_jobs` on Mac OS X 10.9. See GH-636
TST fix sklearn.ensemble.tests.test_bagging.test_parallel
TST move Mac OS checking utility to sklearn.utils.testing
TST skip SparsePCA n_jobs!=1 test on Mac OS X because it hangs
TST remove n_jobs=-1 usages in tests
DOC mention that train_test_split returns a random split
Nantas Nardelli (2):
Added contributing subsection and fixed link to web doc
No whitespaces at the end of file
Nelle Varoquaux (59):
Merge pull request #2324 from arjoly/missing-contributions
DOC added link from banner to example.
DOC now building to html/stable
DOC home always points to stable
Merge pull request #2338 from agramfort/pca_cleanup
ENH added an orange cite us button on the front page
FIX cite us buttong made blue bar span too much
DOC added testimonials
FIX forgot evernote's logo
ENH added telecom to the testimonials
DOC updated evernote's testimonials
ENH added AWeber's testimonial
ENH added carousel back on front page for testimonials
ENH better spacing on the first page
ENH testimonials img are now centered.
FIX typo in testimonials
FIX spelling mistakes and whitespace nitpick
PEP8 fixes on hierarchical.py
ENH improved the documentation of the fix_connectivity function
FIX deprecated the copy arguments in hierarchical clustering
FIX a convert on list was applied twice on the inertia matrix
DOC added the docstring to linkage_tree and AgglomerationClustering
ENH Removed copy option and deprecation on new functions and classes
FIX spelling mistake
MAINT deprecated ward class
TEST test that ran with ward linkage now also run with average and complete linkage
DOC small fixes on the hierarchical clustering
DOC improve narrative docs on hierarchical clustering
TEST FeatureAgglomeration does not behave like normal clustering
ENH AgglomerativeClutering now supports different metrics
DOC/TEST improved failing errors and docstrings on metric
ENH now used instead of euclidean distance to gain speed
DOC/TEST improved doc and tests on the paired distances
FIX spelling mistakes and whitespace nitpick
PEP8 fixes on hierarchical.py
ENH improved the documentation of the fix_connectivity function
FIX deprecated the copy arguments in hierarchical clustering
FIX a convert on list was applied twice on the inertia matrix
DOC added the docstring to linkage_tree and AgglomerationClustering
ENH Removed copy option and deprecation on new functions and classes
FIX spelling mistake
MAINT deprecated ward class
TEST test that ran with ward linkage now also run with average and complete linkage
DOC small fixes on the hierarchical clustering
DOC improve narrative docs on hierarchical clustering
TEST FeatureAgglomeration does not behave like normal clustering
ENH AgglomerativeClutering now supports different metrics
DOC/TEST improved failing errors and docstrings on metric
ENH now used instead of euclidean distance to gain speed
DOC/TEST improved doc and tests on the paired distances
Merge master in hc_linkage
DOC Clarified the doc of the hierarchical clustering
FIX precomputed distances on the hierarchical clustering
ENH callable metrics now work for the hierarchical clustering
FIX the option affinity wasn't used in the hierarchical clustering
DOC metrics and hierarchical clustering
FIX verbosity of the mds
Merge pull request #2831 from eltermann/doc-typo
Merge pull request #2909 from ogrisel/peerindex-testimonial
Noel Dawe (22):
DOC: missing documentation of splitter parameter in tree.py
tree export_graphviz: remove unused close parameter and close the file if out_file is a string
AdaBoostRegressor: fix redundant recalculation of error_vect.max()
plot_adaboost_multiclass.py: handle case where boosting terminated early. Add missing author on other boosting examples.
xrange -> range
add sample_weight to base score and weight_boosting staged_score
weight_boosting: unneeded np.copy
weight_boosting: include sample_weight in test_staged_predict
metrics: add sample_weight support
rm (default=None)
require sample_weight support for binary_metric
newline
atleast_2d.reshape -> reshape
weighted metrics: fix sample_weight handling for average=samples
format
metrics tests
doc: fix default
weighted metrics tests fixes
np.sum(np.multiply( -> np.dot(
update whats_new.rst
add test_base.test_score_sample_weight
sample_weight metrics tests: add missing micro and macro averaging for precision, recall, and f-score
Olivier Grisel (273):
boilerplate
typos
more tutorial boilerplate
missing download script
cosmit
typo
typo
some work on the introductionary section
more work in progress
better title structure and missing paragraph on good feature extractors
more work on machine learning 101
cosmit & typos
wording
missing fetch script for movie reviews
work on exercise 01
better use this a exercise number 2
do not forget to introduce linear separability
typo
add missing fetch script for the face dataset
solution for exercise 04
extracted skeleton for exercise 04
more work on the classifier section
starting explaining PCA
useless conf
more work on PCA
better titles
style
more work on general concepts (esp. supervised learning)
more syle fixes
more subtitles
ignore workspace
ignore OSX stuff
updated the link to the official documentation
work on clustering section
add some subsections separators
wording
reorganizing notes and adding a scikit-learn oriented complement to the supervised learn flow diagram
remove section on density estimation
started work on linearly separable data
added takeway points section
cosmit
various improvements
typo
various cosmits / wording
typo / wording
section on regression models
section on overfitting and the train / test split
removing confusing section + slight reorg
wording
wording
work on text feature extraction doc
one more check in the setup
better intro + more work on text features
more work on text classification example
fix header levels
exercises instructions
improvements in exercises instructions
some practical hints
use the pipeline in exercice 02
inversed solution and skeleton
sentiment analysis
missing skeleton
gh pages integration
better README.rst
use vec.fit_transform instead of vec.transform on the training set
language detection gives slightly better results without IDF
fix broken doctest
Merge branch 'master' of https://github.com/ametaireau/scikit-learn-tutorial into ametaireau-master
Merged pull request #4 from larsmans/master.
Merge pull request #6 from larsmans/master
add the test command
update doctest in tutorial to match current API
fix the random state of the files dataset loader for reproducible results
Upgrade the setup documentation to match the current master
updated README.md
fixed left-over of the previous version
more details for poor windowsians
upgrade to simpler new load_file API
s/class_names/target_names/g
upgraded exercise 2 to the new API
updated faces recognition example to the latest API
upgrade inline doctests to new load_files API
reorder exercises
if __name__ == '__main__' multiprocessing protection in ex2
more details
more details in multiprocessing comment in skeleton too
round numerical results for better stability of the doc tests
use internal dataset fetch + switch to SGDClassifier for faster execution
typo + cosmit
use Perceptron and quick splitter
updgrade to new scikit-learn API
make sklearn version dependency explicit
C is now scaled
working on tutorial exercices skeleton generation
Merge pull request #9 from benjaminwilson/master
Ignore the generated MANIFEST file
Fixed a typo in a contributor's name
Also clean the dist folder when calling make
Add missing credit to @smoitra87 for the Python 3 support
partial_fit for naive Bayes was done for 0.14-rc, not 0.11...
Merge pull request #2348 from arjoly/deprecate-auc_score
DOC: Simpler cross-validation iterator doc
Merge pull request #2369 from larsmans/no-warnings-in-fs
2-fold => 2-fold cross-validation
approximately the same percentage of samples
Remove spurious print statements in sample snippets to make the doc easier to follow
Lowercase y and more consistent blank line usage
Merge pull request #2288 from dengemann/fast_dot
Merge pull request #2396 from dengemann/insert_fast_dot
FIX: renable test_k_means_plus_plus_init_2_jobs on OSX 10.8
Fix broken link
Merge pull request #2370 from ogrisel/doc-cross-validation
Merge pull request #2278 from jnothman/prf_rewrite3
Merge pull request #2222 from FedericoV/Out_of_core_example
More detailed entry for StratifiedKFold fix in whats_new.rst
Revert "More detailed entry for StratifiedKFold fix in whats_new.rst"
Revert "Add entry for #2372 to whats_new.rst"
Revert "Update comment with numbers for when we run with 800 samples."
Revert "Avoid list, preallocate a numpy array for indices instead."
Revert "Instead of linking to NB, explain the problem inside the test itself."
Revert "Fix accidental doctest breakage."
Revert "FIX #2372: StratifiedKFold less impact on the original order of samples."
FIX: more accurate description of eta0 in SGDClassifier
Merge pull request #2442 from glouppe/tree-shuffle
FIX #2372: non-shuffling StratifiedKFold implementation and updated tests
Merge pull request #2463 from ogrisel/stratified-kfold
FIX: skip numpy.dot + multiprocessing test that segfaults under recently updated OSX 10.8
FIX: broken doctest impacted by stratified CV and tree RNG changes combined...
FIX: use next(iterator) built-in instead of iterator.next()
Merge pull request #2495 from jaquesgrobler/ENH_coverage_travis
Merge pull request #2487 from jaquesgrobler/DOC_speed_up_frontpage
Merge pull request #2502 from ericjster/plot_dbscan
FIX: clean-build target does no longer exist
Merge pull request #2516 from jaquesgrobler/add-copybutton-examples
FIX #2481: add warning for bug in old numpy with unicode
Remove leftover print statement
Merge pull request #2519 from ankit-maverick/minor_docfix
Use a RuntimeError instead of a warning to avoid raising a ValueError randomly later
Merge pull request #2523 from ogrisel/skip-numpy-unicode-bug
Release the GIL at tree building time
Put more code under non-GIL block
Merge pull request #2528 from ogrisel/tree-nogil
FIX: make RANSACRegressor pass test_common
FIX: np.ones_like does not support dtype on old numpy
Add a estimator introspection check to test_common
Merge pull request #2538 from ogrisel/test-test-common
Merge pull request #2536 from larsmans/rm-mlcomp-doc-class
Add plot titles and newstyle plt import in OLS vs Ridge example
OPTIM: remove memcopy for X_argsorted in GBRT models
Merge pull request #2524 from rmcgibbo/hmmfix
FIX #1622: OPTIM: remove obsolete random_state instance in the Tree class
Merge branch 'pr/2556'
Better title
ENH: Make forest prediction code more robust to mutations of the estimators list
Merge pull request #2571 from edran/readme
Only run make test when coverage is disabled
Merge pull request #2574 from ogrisel/travis-speed
FIX: errors in kernel parameters for OneClassSVM
Merge pull request #2460 from arjoly/auc-multilabel
FIX: broken sparse matrix check under scipy 0.13.0
FIX: Python 3 dict keys cannot be concatenated with a list
Merge pull request #2631 from pprett/fix-sgd-l1-ratio
ENH: more explicit traceback + message in case of common failure on sparse input
Merge pull request #2592 from jnothman/prf_derivative_warnings
Merge pull request #2649 from jakevdp/polynomial_doc
Merge pull request #2664 from agramfort/fix_randomized_pca
FIX: test failure under windows caused by temp files handling
ENH Upgrade to joblib 0.8.0a2
ENH: reduce memory usage and IPC overhead when fitting forests by using the new threading backend
DOC Update whats_new.rst to document threading backed for forests
FIX: broken sed command under OSX in joblib sync script
ENH: parallize the BaseForest.apply method with the threading backend
Merge pull request #2700 from amueller/refactor_common_tests
Merge pull request #2720 from amueller/chi2_fused_types
ENH: make forests' test_parallel_train run faster
FIX: np.searchsorted numpy bug on unicode objects also impacts 1.6.1
Python 3 compat
Python 3 fixes in exercises
Python 3 fixes for the second exercise
FIX #1565: fix race condition in parallel pre-dispatch by upgrading joblib
Update whats_new.rst for race condition fix
Merge pull request #2756 from ogrisel/joblib-0.8.0a3
FIX #2645: fix 20 newsgroups downloader under Python 3
Add thanks for infrastructure supporters
ENH: made example/svm/plot_iris.py clearer
peerindex testimonial
FIX #2924: make lobpcg test pass with reference and ATLAS impl of LAPACK
More stable test problem for eigen_solver='lobpcg'
Merge pull request #2928 from ogrisel/fix-linalgerror
ENH: update installation instructions for Ubuntu / Debian
Merge pull request #2973 from pprett/datarobot-testimonial
Merge pull request #2959 from Oscarlsson/silhouette_score_label_number
FIX: np.abs might not work on scipy.sparse matrices
Merge pull request #2981 from matrixorz/master
Merge pull request #2997 from GaelVaroquaux/okcupid_testimonial
Merge pull request #2982 from kmike/fix-macos-hangs
FIX: more robust skip of implicit constructor
Merge conflict in whats_new.rst
Merge pull request #3006 from ogrisel/fix-implicit-init-introspection
Merge pull request #3007 from Manoj-Kumar-S/refactor_sparsefuncs
FIX: np.random.randint expects signed 32 bit integers under Windows
FIX: numpy mtrand does not accept Python long instances under Windows
FIX: make python setup.py clean also delete __pycache__ folders
FIX: remove casting warning under Python 3
FIX #3014: use a different folder for covtype data under Python 3
FIX: fix build under Python 2.6
COSMIT: More readable drop for good test
FIX: checked in bad assertion in the last commit by mistake
Remove redundant yet unstable test_spectral_lobpcg_mode
ENH: make CD Lasso raise ConvergenceWarning
Merge pull request #3030 from ogrisel/fix-lars-drop-for-good-test-failure
Merge pull request #3025 from ogrisel/remove-unstable-lobpcg-test
ENH: stable test + catch warning + pep8
ENH: more robust test_toy_bayesian_ridge_object
ENH: configure travis to also test old numpy & scipy
DOC: add some header doc to the travis scripts
More informative numerical error message in SGD
cosmit
Merge pull request #3059 from mdbecker/update_from_33_to_34
Merge pull request #3061 from jess010/ami-docs
Fix merge conflict and missing URL ref in whats_new.rst
Merge pull request #3070 from luispedro/fix-enet-doc
Merge pull request #3062 from abatula/archlinux-install-doc
More Py3 fixes for sphinx build
Merge branch 'python-3-sphinx-fixes'
Merge pull request #3063 from chalmerlowe/improved-digits-example
FIX: One more PY3 fix in the documentation generator
FIX: typo in gen_rst.py
Merge pull request #3069 from ssaeger/issue_3068
Merge pull request #3085 from eickenberg/update_ridge_authors
Merge pull request #3082 from ElDeveloper/install-link
Merge pull request #3067 from mdbecker/truncated_svd_calculate_explained_variance
Merge pull request #3083 from sdenton4/enh_learning_curves
Merge pull request #3086 from mdbecker/update_authors
FIX: remove deprecation warnings in learning curves under Python 3
Merge pull request #3090 from ogrisel/learning-curves-warnings
COSMIT: use plural, there are 2 learning curves
typo
wrap lovely testimonial's paragraph
Merge branch 'pr/3091'
MAINT: ignore coveralls failures
Merge pull request #3145 from ogrisel/travis-coveralls
FIX: use clip(0) instead of abs()
Merge pull request #3182 from arjoly/test-forest
CI: make travis run the doctests
Merge pull request #3189 from ogrisel/travis-doctest
FIX: more Python 3 fixes for doc/gen_rst.py
FIX: euclidean divide in plot_image_denoising.py for Py3 support
PY3: more gen_rst.py fixes for Python 3 compat
DOC: better docstrings for PCA models
FIX: windows test failures in test_ransac
Merge pull request #3169 from mjbommar/issue-3167-eradicate-todense
ENH: PEP257 style + small code simplication
ENH: more explicit failure messages in test_common
MAINT: joblib 0.8.0
Merge pull request #3212 from ogrisel/joblib-0.8.0
Add optional PCA init to t-SNE
MAINT: run tests on files with the exec bit
Merge pull request #3234 from ogrisel/nosetests-exe
MAINT: bump up to scipy 0.14.0 in travis CI config
Merge pull request #3237 from ogrisel/travis-scipy-0.14.0
Merge pull request #3161 from hamsal/ada-sparse
MAINT: joblib 0.8.1
Merge pull request #3242 from ogrisel/joblib-0.8.1
ENH use the np.int dtype to encode integer classes
FIX workaround doctest failure with old numpy
Merge pull request #3248 from MechCoder/remove_precompute_multi
Merge pull request #3246 from ogrisel/rebased-pr-2657
DOC add whats new for precompute fix
Merge branch 'pr/3247'
FIX use NPY_INFINITY instead of C99 INFINITY for MSVC
Merge pull request #3251 from ogrisel/fix-msvc-INFINITY
Branching 0.15.X to prepare the 0.15.0b1 release
Oscar Carlsson (2):
Fix silhouette score n_labels
Added number of labels for debug and regexp in test
Paul Butler (1):
FIX Pipeline should raise ValueError for duplicate name
Paweł Mandera (1):
Fix citation in TfidfTransformer
Peter (1):
cosmit
Peter Prettenhofer (83):
GBRT checks if ``loss`` is in self.supported_loss
renamed supported_loss to _SUPPORTED_LOSS (constants)
fix: typo - y_pred instead of y_true
GBRT enhancements:
added EPSILON_FLT and EPSILON_DBL for almost equal impurity and fx value comparisons
fiddled with EPSILON_DBL
added monitor callback w/ early stopping support
fix: VerboseReporter remaining time in the case of partial_fit
added ZeroEstimator
use StackRecord as elements of stack not 5 consecutive entries
Tree: compute partition specific impurity and total impurity in criterion.children_impurity. Pass partition impurity to stack to avoid re-computation (saves some runtime).
set EPSILON_DBL to old val of 1e-7
add tests for complete
fix: use regressor in regression test
Add stack class
PriorityQueue for best-first tree growing
fix: test renamed and check only on nr of leafs.
tree code now supports both c-style and fortran inputs
tree code now supports both c-style and fortran inputs
updated after tree.pxd change
Tree ensemble classes don't enforce c-style inputs.
use assert_equal if possible
check value of max_leaf_nodes
common test for sparsify + fix for SGDRegressor
sparsify test: add multi-class test too
support 'zero' init, more tests for ZeroEstimator
fix: Huber loss function in gradient_boosting fails if negative_gradient is not called before __call__; now computes on-demand.
add huber loss bug fix to whats new
Merge branch 'master' into gbrt-enh-stackrec-greedy
Implement GBM's best-first heuristic tree growing procedure .
remove asserts
Best-first instead of branch heuristic - identical score as GBM on covertype
cleanup: remove BranchBuilder, added more comments
remove check for condition that might not hold
cosmit
rm MSE import
removed Impurity struct, Criterion.child_impurity now returns impurity_left and impurity_right (not total), fix: moved pos >= end check to if branch
moved data structures (Stack + Heap) to _utils
fix: correct impurity_improvement formula (weight by n_left and n_right)
remove split_impurity from node_split signature
update test (slight changes in tree output because of removal of EPSILON insensitive checks)
use max_leaf_nodes in regularization example
narrative documentation on controlling tree size - might be relevant for weighted boosting and CART as well.
rename GBM_MSE to FriedmanMSE
prefer np.as(fortran|continuous)array over np.asarray(order=order)
fix: pass warm_start to BaseSGDRegressor
Merge pull request #2617 from pprett/sgd-regressor-fix
pass impurity to ``impurity_improvement`` - now it is correct so that arbitrary Criteria can be used in best-first search.
Merge pull request #7 from glouppe/gbm
X need not be continous anymore (since we adhere to col and row strides)
cosmit: doc undocumented functions
cosmit: remove commented out code - add comment
Merge pull request #8 from glouppe/gbm
max_leaf_nodes has precedence over max_depth if the former is not None.
fix: l1_ratio is incorrect; its (1.0 - rho); added test case to make sure elastic net penalty with l1_ratio near to 1 and 0 matches L1 and L2 penalties.
added change log entry
set l1_ratio to (1.0 - l1_ratio)
revert: reverted l1_ratio change - not SGDRegressor - sorry
Merge branch 'master' into gbrt-enh-stackrec-greedy
Merge pull request #9 from glouppe/gbm
Merge branch 'master' into gbrt-enh-stackrec-greedy
renamed partial_fit -> fit_more for GBRT
moved fit_more to warm_start=True
fix: resize oob improvements, train score when early stopping
warm_start semantics now fit exactly n_estimators rather than self.estimators_ + self.n_estimators
fix: use estimators_.shape[0] rather than n_estimators to make predictions; what if user just changes n_estimators of an already fitted obj.
fix: warm_start demo in narrative documentation
updated whats new
malloc / realloc checks
cosmit: better errors
do not alter n_estimators after early stopping
cosmit: better exception
cosmit: better exceptions
fix: wrong placeholder in format string
better test coverage in tree module
GBRT subsection for warm_start
Merge pull request #2570 from pprett/gbrt-enh-stackrec-greedy
fix: remove print from gbrt test
fix: use six.string_types instead of basestring
uses raises instead of with assert_raises - apparently doesn't work on python3 nose version.
fix: remaining time not currectly computed
Merge pull request #2753 from pprett/gbrt-verbose-remaining-time
add datarobot testimonial
Philippe Gervais (21):
Added euclidean_distances_argmin
Chunking on both arrays in euclidean_distances_argmin
Improved tests for euclidean_distances_argmin
Wrote docstring for euclidean_distances_argmin
[BUG] safe_asarray() converts sparse matrices dtype
Used safe_asrray() in check_pairwise_arrays()
Change signature of euclidean_distances_argmin()
Added pairwise_distances_argmin()
Optimized euclidean case in pairwise_distances_argmin
Removed euclidean_distances_argmin
Minor code cleanup
Added gen_batches() to utils
Changed pairwise_distances_argmin API
Code cleanup in pairwise_distances_argmin
[DOC] Fixed pairwise_distances_argmin docstring
[API] Renamed pairwise_distances_argmin
[DOC] updated pairwise_distances_argmin_min doc
Performance enhancement for non-euclidean metrics
Small fixes
CSR matrix support in pairwise_distances_argmin_min
Added argument metric_kwargs
Pietro Zambelli (1):
FIX unused pos_label parameter in metrics.precision_recall_curve
Rajat Khanduja (5):
Updated examples to use pyplot for plotting instead of pylab
Fixed pep8 violations. Some 'line too long' errors still remain.
Fixed some PEP8 violations even present in master branch, in examples/
Some more examples updated to use matplotlib.pyplot
Modified example 'plot_stock_market.py' to use matplotlib.pyplot
Ralf Gommers (1):
MAINT: fix broken links to numfocus.org on donations page.
Robert Layton (4):
Update to mean shit clustering narrative documentation.
Update to docstring of meanshift
docstring of module
Fix typos found by Alexandre
Robert McGibbon (5):
Fix bug identified in #1817, comment 17340049
Fixes to HMM docstrings
Clarification of the sequence length, per @ogrisel
Add inline comment
Change the order of test_score and train_score in the _fit_and_score docstring to reflect what the code actually does
Rohit Sivaprasad (2):
Edit typo
DOC typo in SVM narrative
Roland Szabo (2):
[DOC] Remove that RBM's are not implemented yet
[DOC] Typo fixes in documentation for Novelty and Outlier Detection
Ronald Phlypo (10):
bug fixes in graph_lasso and moving non-custom log-likelihood definition to specific definition of cost
corrected sign error in _objective
pep8 compliance
added test cases to test_graph_lasso for alpha=0
deleted the .orig file in this commit
deleted *.py.orig from branch
cosmetic remarks incorporated
Merge branch 'glbugfixes'
cosmetic remarks, pep8 compliance against trailing white spaces, spelling errors
clean up
Ronan Amicel (1):
Minor edits.
Rupesh Kumar Srivastava (1):
FIX max_features in CountVectorizer
Samuel St-Jean (1):
Update dict_learning.py
Sebastian Saeger (4):
FIX: AdditiveChi2Sample can be initialized with sample_interval, #3068
Add tests for the sample_interval in #3069
Fix the pep8 violations
TST MiniBatchKMeans with many reassignments
Sergio Pascual (5):
Remove execute pemission
Remove shebang in library code
Move __future__ import after license text
MAINT Update six to version 1.4.1
Add a paragraph about installing a binary release in Fedora
Simon Frid (2):
Update testimonials.rst
adding lovely logo
Skipper Seabold (1):
ENH: Raise explicitly on non-unique vocab.
Stefan Walk (1):
python 3 compatibility fix
Stephan Hoyer (1):
Update permutation_test_score docstring
Steven Maude (3):
Update cross_validation.rst
Update feature_selection.rst
Minor typo fixes in grid_search_text_feature_extraction.py
Thomas Unterthiner (8):
Issue #2455: Make RBM work with sparse input when verbose=True.
Change RBM sparse format from CSC to CSR.
Cosmetic changes.
Add issparse import.
Cosmetic changes.
Expanded docstring of the verbose parameter.
Check that verbose output is sound.
Extend utils.sparsefuncs: inplace scale and axis min/max
Vijay Ramesh (1):
adding change.org testimonial, logo
Virgile Fritsch (4):
BF: Correct degrees of freedom in f_regression + test.
BF in f_regression: variable naming + use assert_*array*_almost_equal.
f_regression and degress of freedom: update whats_new + minor.
Merge pull request #2960 from ethanwhite/docs-typo
Vlad Niculae (2):
Add .bmp sklearn logo of correct width for Win setup
[TYPO] s/migh/might/
Will Myers (1):
Fixed SelectKBest corner case: k=0
Yaroslav Halchenko (6):
BF: use hasattr with providing attr name (Thanks to Timo Schulz)
Merge tag '0.15.0b1' into releases
Merge branch 'releases' into dfsg
Merge branch 'dfsg' into debian
changelog, refreshed patches
Build-depend on libatlas3-base
Yoni Ben-Meshulam (1):
Fix a minor typo: 'They requires' should be 'They require'.
Yoshiki Vázquez Baeza (2):
DOC: Remove dead link
DOC: Change URL-based link to sphinx :ref:
Yung Siang Liau (2):
FIX: Add allow_nans option to check_arrays
FIX TfidfVectorizer exports idf_ attribute
Yury V. Zaytsev (1):
BUG: typo fixes in sklearn.mixture.gmm
abhishek thakur (1):
staged_predict predicts classes, not probabilites
adrinjalali (1):
svm fit numpy array indexing deprecation warning fix.
ai8rahim (1):
Changed assert_array_equal() in Line 45 and 46 to assert_array_almost_equal(,,decimal=5). This has fixed the AssertionError, which occurs during the installation test.
alemagnani (1):
FIX+TST non-consecutive or duplicate vocabulary indices
amormachine (2):
Updating copyright year to 2014
Update AUTHORS.rst
benjamin wilson (1):
remove reference to removed API, fixes #6
bwignall (8):
Add option to restrict LassoCV to positive-only coefficients
Add option to restrict ElasticNet to positive-only coefficients; add test for this case, and the same for LassoCV
Fix some typos in whats_new.rst, using ispell
MAINT: Fix PEP8 warnings in sklearn/covariance
DOC: Replace GT/LT with angle brackets for inner product
CLN: Fix typo
CLN: Capitalize "Gaussian" in example docstrings
CLN: Capitalize "Dirichlet" and "Mexican" in example docstrings
cgohlke (3):
Fix ValueError: Buffer dtype mismatch, expected 'INTP' but got 'long' on win-amd64
FIX MSVC compile error C2036: 'void *' : unknown size
TST: Fix ValueError: Buffer dtype mismatch, expected 'npy_intp' but got 'long' on win-amd64
chalmerlowe (1):
Improved plot digits classification example.
cjlin (3):
MAINT import LibSVM 310
add a comment for sigmoid_predict in svm.cpp
MAINT import LibSVM patch from upstream
csytracy (2):
FIX "ValueError: startprob must sum to 1.0" in HMMs
FIX stability in HMMs
dzikie drożdże (1):
PY3 file handling
eltermann (8):
Small documentation fix: from CMS to VCS
Fixed documentation typo
Fixed documentation on sklearn/tree
Replaced abbreviated 'w.r.t' to 'with regards to'
Fixed small typo
Doc fix - Compiled .pyx files with Cython 0.20
/s/2013/2014: Updated project copyright date
s/svn/svm
hamzeh (3):
Implemented a check for ndim exceeding two in the utils.check_arrays function
Removed unnecessary variable n_dim in utils.check_arrays, removed unnecessary parens in same place
Trivial change in utils.check_arrays from > 2 to >=3 in attempt to rebuild Travis CI
jamestwebber (1):
DOC: removed duplicated Python 3 section
jess010 (2):
updated AMI documentation for issue #2686
update ami doc for feedback on PR
john collins (3):
LOO is bad doc
updated for typos and a few wording changes
one last comment fix
kaushik94 (3):
Update supervised.py
Update supervised.py
ENH add sparse parameter to OneHotEncoder
kowalski87 (2):
fixed bug in mds.py
added not
matrixorz (1):
fix issue #2901
murad (3):
added tests for sparse matrix inputs to BaggingClassifier and BaggingRegressor
open iris files with statement to avoid ResourceWarning
BaggingClassifer/BaggingRegressor tests for sparse input
popo (1):
MAINT some more pointer safety in LibSVM
samuela (1):
Fix typo in the Gaussian PDF
sdenton4 (4):
FIX: More robust test_extmath.py, compatible with older numpy
ENH: Compute lower number of points by default
Improved the learning curve example.
Fixed style errors detected by pep8.
staubda (1):
Improved documentation of the "estimator_params" argument for RFE and RFECV.
t-aft (1):
DOC grid_search_digits.py does no do nested CV
ugurthemaster (6):
Update plot_digits_pipe.py
Update plot_tree_regression_multioutput.py
Update plot_forest_importances.py
Update plot_svm_regression.py
Update working_with_text_data.rst
Update plot_adaboost_regression.py
-----------------------------------------------------------------------
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