[scikit-learn] annotated tag 0.4 created (now 1be8e13)
Andreas Tille
tille at debian.org
Wed Dec 28 13:10:54 UTC 2016
This is an automated email from the git hooks/post-receive script.
tille pushed a change to annotated tag 0.4
in repository scikit-learn.
at 1be8e13 (tag)
tagging 65d06f830ec6604b44d1a0510255868a8f762e3a (commit)
replaces 0.2
tagged by Fabian Pedregosa
on Thu Aug 26 13:10:52 2010 +0200
- Log -----------------------------------------------------------------
0.4 final release
Alexandre Gramfort (57):
fix doctest in crossval
adding UnivSelect object in univ_selection to by pass the use of predictor object. The UnivSelect implements fit + reduce rather than fit + predict
fix univ_selection test
adding leave_p_out in crossval.py
allowing Lasso CD + Elastic-Net CD to pickle
adding LDA implementation with test + example
adding "proba_predict" method in LDA (should be done for SVC too)
fix docstring test in Neighbors class
cosmit in naive_bayes.py
cosmit in lda
correct typo
uniforming the access to the primal variables (the "w") with a "coef_" property
SVM : renaming coef_ to dual_coef_ in SVM
Logistic : adding properties to get access to coef_ and intercept_
improve computation of coef_ in SVC
adding doc to LogisticRegression class
update in doc for LogisticRegression
introducing intercept_ in SVM (from libsvm)
adding example to view posterior class probabilities when using LogisticRegression or SVC
adding example that illustrates that L1 penalty leads to sparse estimates of coef_
adding docstring to LinearSVC and test to make sure the coef_ and intercept_ returned by SVC and LinearSVC are the same.
adding predict_proba method in UnivSelection class
Merge branch 'coordinate_descent_2' of http://github.com/fseoane/scikit-learn
Merge branch 'coordinate_descent_3' of http://github.com/fseoane/scikit-learn into HEAD
changing interface of lasso + enet for consistency with GLMNET
bug fix in coordinate descent for GLMNET
changing interface of lasso + enet for consistency with GLMNET
bug fix in coordinate descent for GLMNET
changing interface of lasso + enet for consistency with GLMNET
bug fix in coordinate descent for GLMNET
merge origin
refactoring to avoid writing model_params in path object construction
removing unnecessary imports
adding __str__ method in LinearModel. adding examples for lasso + elastic + path with crossval. adding computation of explained variance (ie. r^2). disabling early stopping in paths as it turns out on an example that it tends to stop the computation too early
bug fix in elastic-net dual gap computation
fix docstring test in cross_val LeaveOneLabelOut
adding __str__ method in LinearModel. adding examples for lasso + elastic + path with crossval. adding computation of explained variance (ie. r^2). disabling early stopping in paths as it turns out on an example that it tends to stop the computation too early
adding possibility to recover the length ie the number of folds of a cross-validation object
fix typo
Merge branch 'master' of ssh://scikit-learn.git.sourceforge.net/gitroot/scikit-learn/scikit-learn
ENH: adding possibility to compute confusion matrix
ENH: adding possibility to compute ROC curve
ENH: adding possibility to compute precision-recall curve
moving ROC etc. to metrics.py
Merge branch 'master' of ssh://scikit-learn.git.sourceforge.net/gitroot/scikit-learn/scikit-learn
cleaning up my mess...
adding references to wikipedia pages in metrics.py
Merge branch 'master' of ssh://scikit-learn.git.sourceforge.net/gitroot/scikit-learn/scikit-learn
updating ROC example and adding an ROC example with cross-validation
BUG: fix in precision recall example
BUG: fix in test metrics
API: changing the way the parameters of Lasso+E-Net are optimized
ENH : imroving documentation of lasso + enet paths function
typos
ENH: new univariate feature selection code
cleaning plot_lar example
API: renaming LassoPath to LassoCV and ElasticNetPath to ElasticNetCV
Chris Filo Gorgolewski (1):
calculate likelihood of the GMM model in the log dimension
Fabian Pedregosa (204):
Start 0.3 development cycle
Fix import path in BallTree benchmarks
Add another balltree benchmark, this time with plot interface.
Add a test for OneClassSVN.
Fix docstring for svm.SVC
Added svnauthors file
Remove mailmap file
Add feature_selection to setup.py
Remove empty directories
Remove old datasets.
Remove empty directories
Revert "RF: build Cython code during build time"
Revert "RF: removing Cythoned C code"
Preliminary sphinx example generation
Add directory for auto-generated examples.
More on auto-generated examples
Temporary hack to generate the examples.
Generate examples dinamically.
Update documentation.
Cosmetic changes to gen_rst
Fix typos in examples.
DOC: howto upload generated doc to sf.
Update README
Preliminary Logistic regression classification using liblinear.
Implent L1-penalized Logistic Regression.
Some bug fixes for liblinear bindings.
BUG: More bug fixing around liblinear bindings.
ENH: Manage special case of 2 classes in LogisticRegression.
Preliminary libsvm support of probability predict.
Preliminary support for probability estimate in SVM.
Turn off probability estimates by default.
Implement a test for SVC probability estimates.
Bug fixing in SVM predict probabilities.
Use liblinear in SVM module when kernel is linear.
BUG: Fix bugs in liblinear bindings.
Remove liblinear-related code from SVC.
Implement LibSVC class for liblinear bindings.
Implement probability predict in Logistic and LinearSVC
Implement probability predict in the case of l1-logistic regression.
DOC: Specify git repo in development section.
Remove empty directories
Fix compatibity issues of bindings in 64bit.
Always use np.float64 for compatibility.
Use syste-wide BLAS libraries if available.
Update svm benchmarks
Fix typo in setup.py
Refactoring on the svm module.
ii
Add more tests to test_svm.py
Add the option to link against system-wide libsvm by editing site.cfg
Remove unused inports from lasso_cd
Add logos (in svg and bmp format)
Remove examples from module glm.coordinate_descent.
FIX: Compatibility Visual Studio
FIX: Fix a bug that caused segfault when intercept=False on classes
Do not execute tests in degenerate cases.
0.3 Beta release
Add download link to main web page.
ENH: add docstrings to module svm
Remove scikits.optimization as requirement.
0.3 release
Start of 0.4 development cycle.
Update information relative to git repo in README
Typo in README.
Merge branch 'master' of ssh://scikit-learn.git.sourceforge.net/gitroot/scikit-learn/scikit-learn
Refactor lasso coordinate descent code.
More work on coordinate descent
more work
more work.
more work
finally working
Update repository location in documentation.
more
More on updating the doc.
Add also a link to the repo via ssh
more on coordinate descent
Fix broken setup.py in datasets.
Refactor lasso coordinate descent code.
Add test directory to setup.py and update dependencies on README.
Updates and fixes for the svm documentation.
Refactor lasso coordinate descent code.
Update references to the svn in the docs.
Implement predict_margin in classes that wrap libsvm.
Remove references to svn repo.
Refactor lasso coordinate descent code.
ENH: documentation for coordinate_descent
Add generated cython code for cd_fast.
Implement predict_margin in classes that wrap libsvm.
Add tests for predict_margin.
BUG: fix bug in tutorial
Add support for precomputed kernel.
BUG: fix bug in automatic doc generator.
Update doc.
Update test after solved bug in elastic net path.
A bit of documentation on svm precomputed kernels.
Welcome Chris.
Update docstrings in svm module.
Add a complete example of SVM classifier using a precomputed kernel.
Update examples.
Give appropriate substitutes for modules not in python 2.5.
Updates for the doc.
More documentation from BallTree object into sphinx documentation system.
Add some notes on how to run the development version.
Remove svn_ignores
Welcome Angel Soler.
Fixes in Ordinary Least Squares (glm module)
Refactoring: Squash samples_generator module into one file.
More docstrings for module glm.regression.
Add an optional argument intercept to Ordinary Least Squares.
Remove outdated script.
Fix broken link in rst feature_selection.
Remove unused parameters from objects that depend on libsvm and liblinear.
Removed unused doc.
DOC: more documentation for svm
Implement class weights in classes that query libsvm.
Fix imports in glm test
Cometic: Change name of weights parameter in Libsvm.fit
Simplify example svm_hyperplane
Add an exmple of svm using weighted classes
Rename plotted examples.
DOC: fixes and enhacements for the docs.
Add title to examples
FIX: fixes in svm predict_proba
Cosmetic changes to the svm tests.
Initial import of Ron Weiss' gmm module.
Add intercept to classes Lasso and ElasticNet
Cosmetic changes in SVM doc.
Fix failing tests because of added intercept.
Remove property _ndim (replaced with field ndim) in gmm
Rename train --> fit in gmm
Welcome Ron Weiss.
Some docs on contributing to the project.
Doc and cosmetic changes in coordinate_descent.
Refactoring: squash glm module in a single file.
Initial import of LAR algorithm.
Fix blas build for liblinear when no system-wide blas is found.
Add .mailmap file.
Fixes for LARS algrithm.
Refactoring in glm module.
Remove logging from gmm module.
Add disclaimer and some comments to gmm module.
Refactoring in gmm module.
Update API docs.
Update contribute doc.
Update doc in glm module.
Some fixes for LARS
Fixes for LARS algorithm.
Remove duplicated files.
Add compute values of lambdas in LeastAngle Path.
Some fixes for LAR and added example.
Cosmetic changes to LeastAngleRegression.
Cosmetic changes in LAR.
Add the diabetes dataset and modify the LAR example.
LAR segfault fixes and cleanup.
Update benchmark to stay in sync with pymvpa developement.
Typo in glm.
Fixes for lars.c to compile under Visual Studio.
LeastAngleRegression fixes & refactoring.
Call gc explicitly in the benchmarks to avoid unnecessary noise.
Initial import of shortcuts module.
Cleanup in LAR example
GMM refactoring.
Example on plotting a gaussian mixture.
Remove manifold module.
Update tests (fix broken tests on 64 bits).
Remove the shortcuts module.
Rename libsvm --> _libsvm
Rename module BallTree -> ball_tree
Use np.allclose in gmm instead of custom functions.
Remove outdated doc.
Update doc with new gmm module.
Normalize in LAR fit
FIX a test.
FIX: typo in setup.py
Rename C extension liblinear -> _liblinear
Rename minilearn -> _minilearn
More doc on contribute.
Add reference to mailing list in main page.
Refactoring in the glm module.
Add more tests for LeastAngleRegression.
Small modification in the LARS algorithm.
FIX: intercept in ElasticNet
Use default values in lasso example.
GLM refactoring: put explained_variance_ as a public field.
Some fixes to make it compile against MKL library.
Updates on gmm example: cleanup.
Update example gmm.
Fix broken links in doc.
Add an example with probability distribution estimates using GMM.
Remove examples from deprecated modules.
Remove duplicate module univariate_selection.
Remove old em module.
Initialize with a fixed seed tests in feature_select
Remove redundant seed generator.
Add docstring to UnivariateFilter.
Comment out broken parts of the feature selection example that are broken.
Add a __version__ attribute.
Remove also auto-generated examples with make clean
Rename examples that do not plot results.
Move glm examples to common directory.
Remove benchmarking code from lasso example.
Add title to example.
Remove code from roc plotting that was interfering with build system.
0.4 release.
Gael Varoquaux (43):
MISC: Safety commit
BUG: fix for ticket 39: selecting 100% should now select 100%.
MISC: Set svn ignores.
DOC: Improve the feature selection example.
DOC: Small changes to the documentation building to be able to references
DOC: Sprucing up the documentation.
BUG/DOC Correcting bugs in the documentation generation code.
Workaround for proxy problems.
DOC: Adding plotting to feature selection, in 2 times to work around
MISC: Cosmetic commit: lines longer than 80 chars.
ENH: Refactor the datasets to add the digits standard dataset.
MISC: Cosmit (cosmetic commit): s/label/labels, because it is a list.
BUG: Proper description for digit dataset.
BUG: Fix the digits datasets: wrong data loading.
MISC: Remove the checked-in generated documentation files from the
BUG: Correct bug in digits dataset loading introduced by previous
BUG: Make sure the docs build even if the directories do not exist.
ENH/DOC: Add an example doing classification on digits.
DOC: Update docs README to use rsync and try to be careful about the
ENH: Make it possible for SVM.predict to work on a single sample.
DOC: Improve the example docstrings.
DOC: Work on the front page.
Improve front page
DOC: Some work on the beginning of the doc.
DOC/MISC: correct docstring on digits loading code.
DOC: Adapt the getting started tutorial to complete beginners.
DOC: Moving the em examples. Doing this in two times, to avoid what looks
DOC: Moving the em examples, step 2.
DOC: Rework of the docs and examples: multiple commits because commits
DOC: Moving SVM examples in a separate folder, step 1.
DOC: Moving examples in a separate directory, final commit (hopefully)
MISC: Typo.
ENH/API: Change the cross-validation utilities from generators to objects
MISC: Rename *_indexes to *_indices in cross_val.
API: Remove the obsolate attrselect.py module.
Removing code in __main__ for univ_selection
Cosmit
Enhance plot_classification_probability example.
Merge branch 'master' of ssh://gvaroquaux@scikit-learn.git.sourceforge.net/gitroot/scikit-learn/scikit-learn
Fix import path in examples.
Beautify the SVM docs
Merge branch 'master' of ssh://gvaroquaux@scikit-learn.git.sourceforge.net/gitroot/scikit-learn/scikit-learn
More beautify
Olivier Grisel (36):
s/n_features/n_samples/ in digits example
add some print statements to monitor lasso/enet progress and tweaks parameters to make it faster to run
Merge branches 'master' and 'master' of git://github.com/yarikoptic/scikit-learn
Merge branches 'master' and 'master' of git://scikit-learn.git.sourceforge.net/gitroot/scikit-learn/scikit-learn
typo
.gitignore: vim power! + generated __config__.py and build
Merge branches 'master' and 'master' of ssh://ogrisel@scikit-learn.git.sourceforge.net/gitroot/scikit-learn/scikit-learn
rename README to README.rst so that it looks nice on github
include the cython generated C code for cd_fast
Merge branches 'master' and 'master' of ssh://ogrisel@scikit-learn.git.sourceforge.net/gitroot/scikit-learn/scikit-learn
Merge branch 'coordinate_descent_4' of http://github.com/fseoane/scikit-learn into coordinate_descent_4
fixed tests
PEP8 is beautiful
include the cython generated C code for cd_fast
yet another attempt at early stopping with regularization path on LASSO + CD
note for later
merging from master
ENH to LassoPath: cleaner, shorter code and storing path is now optional
Merge branch 'master' of ssh://ogrisel@scikit-learn.git.sourceforge.net/gitroot/scikit-learn/scikit-learn
Merge branch 'master' into ticket-51-cd-early-stopping_2
removing wrongully added conflict markers
make fit work on python tuples and lists
spare one local variable
oriented programming magic to factorize common logic between Lasso and Elastic Net
some tests for the elastic net path-based solver
fix name conflict issue with stdlib math module
make the LogisticRegression fit return the classifier instance for chained calls
make the LinearSVC fit return the classifier instance for chained calls
make loss and penalty notation casing consistent accross scikit.learn (LinearSVC and LogisticRegression)
merge master heads
cleanup conflict marker leftover in test_svm.py docstring
FIX: make the window_size parameter taken into account in the BallTree implementation of KNN
LARS test clean-up
fix module path in pickled samples for manifold learning
add note for IRC chan
better tests for LeastAngleRegression
Vincent Michel (7):
Fix a bug when using an estimator in UnivSelection.
Add a function for crossval : leave-one-label-out
Add Gaussian Naive Bayes classifier
Fix a bug in the computation of the log lokelihood
Add coef_ to SVR.
New univariate feature selection.
Add support function for computing the selected features.
Yaroslav Halchenko (16):
DOC: minor spelling mistakes (just to check upload rights and my login)
adding myself to svnauthors
DOC: extending README with information about GIT Mirror
DOC: minor changes in README in regard to use of git-svn
BUG: removing evil tab
BUG: removing add_subpackage for deprecated datasets
RF: build Cython code during build time
RF: removing Cythoned C code
RF: removing obsolete .pxd files
BUG: correct import within test_densities.py
RF: use assertEqual to compare str(e) instead of a print (could simply be assertRaises)
BF: enforce little-endian byte order for loadmat call in test_gmm_em (Closes: #580879)
Adding myself into AUTHORS and fixing up atypo
BUG - LDA, separate out "priors" possibly estimated from the ones specified in constructor
extending LDA unittest to catch issue fixed with previous commit + 1 more (just in case)
pylint friendliness -- spaces before commas, use warnings, etc
-----------------------------------------------------------------------
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