* [MRG + 1] Fix failure on numpy master (#8011)

Was causing "ValueError: The truth value of an array with more than one element is ambiguous"

* [MRG+1] Add multiplicative-update solver in NMF, with all beta-divergence (#5295)

* FIX .format arguments were in the wrong order

Add check_no_fit_attributes_set_in_init test and use name in the error
message rather than estimator since the former is more readable.

* left-over deprecation of 1d X (#8045)

* [MRG + 1] CI some improvements to the flake8 CI (#8036)

Examples now cannot fail due to 'E402 module level import not at top of file'

* [MRG] Set min_impurity_split in gradient boosting models (#8007)


Fixes #8006

* Use 1.0 not 1 in error message regarding float value

* DOC add CI details and commands to contributor guide (#8024)

* DOC Update LOF.fit_predict() (#8059)

Changed documentation to remove a self contradiction.

* TST fix test case which should ensure empty row (#8056)

* [MRG+2] ENH add n_jobs to make_union through kwargs (#8031)

*  DOC adding note regarding bessel correction in PCA (#7843)

* Fix plot_svm_margin example plots (#8051)

* Corrected shift of separating hyperplane that passes through the support vectors

* Fixed signs on yy_down and yy_up

* PEP 8 spacing correction

* Fixed spaces at end of comment for Travis

* DOC fix broken link in carousel

Fix #8081

* [MRG + 1] Reformat the version info and cite us labels in the user-guide docs. (#8072)

* Reformat the version info and cite us labels in the user-guide docs.

* Fix the line break in the version number.
* Refactor the cite us sentence to have cite us in the beginning of
  the line.

* Edit the sentence formation of "cite us" section.

* [MRG + 1] Fix reference in fetch_kddcup99 (#8071)

* fix 'percent10' parameter default in fetch_kddcup99 docstring
* Consistent default 'percent10' value in _fetch_brute_kddcup99 to be consistent

* [MRG + 1] Issue#8062: JoblibException thrown when passing "fit_params={'sample_… (#8068)

* Issue#8062: JoblibException thrown when passing "fit_params={'sample_weights': weights}" to RandomizedSearchCV with RandomForestClassifier

* Added test for issues #8068 and #8064.

* Clean up with pyflakes.

* Changed cryptic comment.

* [MRG + 1] Fix perplexity method by adding _unnormalized_transform method, Issue #7954 (#7992)

Also deprecate doc_topic_distr argument in perplexity method

* [MRG+1] allow callable kernels in cross-validation (#8005)

* DOC Fix doc for CountVectorizer class. (#8085)

* Change the return type mentioned in the comments from
  scipy.sparse.coo_matrix to scipy.spartse.csr_matrix.

* DOC clarify logisticregression n_jobs param (#8083)

* CI fix bug in getting changed docs when no sklearn/ files modified

* DOC Document _changed.html in contrib docs

* DOC Restructure the version info in the docs to fit in two lines. (#8084)

* FIX check_array's accept_sparse param now takes true/false/str/list, but not None (#7937)

* DOC Fix output shape in doc for OrthogonalMatchingPursuit (#8091)

* [MRG + 2] Allow f_regression to accept a sparse matrix with centering (#8065)

* Updated centering for f_regression

Allows f_regression to accept a sparse matrix when centering=True.

* Fixed E226 spacing issue.

* Added f_regression sparse update to whats_new.rst

* DOC Improve benchmark on NMF (#5779)

* ENH improve benchmark on nmf

* add projected gradient solver inside the benchmark file

* add comments and authors for _PGNMF

* CI limit diff to commit range in flake8_diff.sh (#8097)

* DOC: Fix the documentation of scoring LogisticCV (#8099)

* [MRG+1] Corrected sign error in QuantileLossFunction (#6429)

* Corrected sign error in QuantileLossFunction

* [MRG+1] Return list instead of 3d array for MultiOutputClassifier.predict_proba (#8095)

* Return list instead of 3d array for MultiOutputClassifier.predict_proba

* Update flake8, docstring, variable name

 - Changed `rs` to `rng` to follow convention.
 - Made sure changes were flake8 approved
 - Add `\` to continue docstring for `predict_proba` return value.

* Sub random.choice for np.random.choice

`np.random.choice` isn’t available in Numpy 1.6, so opt for the Python
version instead.

* Make test labels deterministic

* Remove hanging chad...

* Add bug fix and API change to whats new

* [MRG + 1] Add changelog entry for MSLE implemented in #7655. (#8104)

* DOC fix link in what's new

* DOC Note how ariddell/lda differs from sckit-learn's LDA (#5553)

Fixes #5529

* COSMIT PEP257

* [MRG + 1] MAINT Move heapify_up/heapify_down into PriorityHeap as class methods + COSMITs (#7034)

* RFC: move heap methods to class and remove trailing spaces

* spurious comment to force recythonization of boosting

* [ci skip] remove spurious comment

* remove trailing whitespace on line

* style: fix trailing whitespace in _criterion.pxd

* add spurious comments to try to force recythonizing

* remove changes for recythonization

* DOC Fix help link on about page (#8119)

* [MRG+2] FIX IsolationForest(max_features=0.8).predict(X) fails input validation (#5757)

Fixes #5732

* DOC Fix indentation errors and username links (#8121)

* [MRG] MAINT Python 3.6 fixes (#8123)

* FIX dict order dependent doctest

* FIX str concat TypeError message changed in Python 3.6

* MAINT upgrade travis config with most recent conda deps

* [MRG+3] Fused types for MultiTaskElasticNet (#8061)

* Convert ElasticNet Multioutput to floating.

* Remove all the float64 coordinate_descent.

* Add the necessary cblas for use fused types.

* Fix zeros dtype issue in cd_fast.

* Remove cblas files.

* Change random seed to let test_lle_simple_grid pass.

* Add tests to check floatting issue for MultiTaskElasticNet.

* Update cblas_sscal.c

* DOC add sklearn-crfsuite to related projects (#7878)

* [MRG+1] Catch cases for different class size in MLPClassifier with warm start (#7976)  (#8035)

* added test that fails

* generate standard value error for different class size

* moved num_classes one class down

* fixed over-indented lines

* standard error occurs a layer up.

* created a different label comparison for warm_start

* spaces around multiplication sign.

* reworded error and added another edge case.

* fixed pep8 violation

* make test shorter

* updated ignore warning

* FIX Split data using _safe_split in _permutaion_test_score (#5697)

Squashed commits:
[94fd9f4] split data using _safe_split in _permutaion_test_scorer
[522053b] adding test case test_permutation_test_score_pandas() to check if permutation_test_score plays nice with pandas dataframe/series
[21b23ce] running test_permutation_test_score_pandas on iris data to prevent warnings.
[15a48bf] adding safe_indexing to _shuffle function
[9ea5c9e] adding test case test_permutation_test_score_pandas() to check if permutation_test_score plays nice with pandas dataframe/series
[3cf5e8f] split  data using _safe_split in _permutaion_test_scorer to fix error when using Pandas DataFrame/Series

* DOC Fix typo in FAQ (#8132)

* [MRG] update copyright years for 2017 (#8138)

* [MRG+1] Fix "cite us" link in sidebar (#8142)

* Fix citation link in sidebar

* Improve formatting of 'cite us' link

* [MRG+1] Add DBSCAN support for additional metric params (#8139)

* Add DBSCAN support for additional metric params

* [MRG+1] fowlkes_mallows_score: more unit tests (Fixes #8101) (#8140)

* DOC: updating GridSearchCV's n_jobs parameter (#8106)

fixes #5797

* [MRG+1] Deprecate ridge_alpha param on SparsePCA.transform() (#8137)

* FIX sphinx gallery rendering of plot_digits_pipe example

* [MRG+1] DOC: complete list of online learners (#8152)

* DOC: complete list of online learners

I have purposely left out the RBM, because it is a model that has fallen
out of fashion and should probably not be used.

* DOC: Add MaxAbsScaler to online learners

* [MRG+2] Avoid failure in first iteration of RANSAC regression (#7914)

Fixes #7908 

Adds RANSACRegressor attributes n_skips_* for diagnostics

* [MRG] FIX Avoid default mutable argument in constructor of AgglomerativeClustering (#8153)

* [MRG + 1] add partial_fit to multioutput module (#8054)

* add partial_fit to multioupt module

* fix range in python3

* fix flake8

* fix the comments

* fix according to comments

* fix lint

* remove pytest

* fix ValueException message

* py 3.5 compatiable classes

* fix stuff

* fix according the comments

* remove used copy

* flake8..

* fix docs

* eventually, i use deepcopy to ensure the parallel

* lint..

* address final comment

* fix addressing the comments

* update confirmed separate estimators

* finally remove copy

* compact test

* [MRG + 1] Add fowlkess-mallows and other supervised cluster metrics to SCORERS dict so it can be used in hyper-param search (#8117)

* Add supervised cluster metrics to metrics.scorers

* Add all the supervised cluster metrics to the tests

* Add test for fowlkes_mallows_score in unsupervised grid search

* COSMIT: Clarify comment on CLUSTER_SCORERS

* Fix doctest

* Fix Ridge floating point instability (#8154)

* DOC Fix link (#8171)

* [MRG + 1] Fix the cross_val_predict function for method='predict_proba' (#7889)

Handle the case where different CV splits have different sets of classes present.

* fixing typo in cs_mse_path_ deprecation (#8176)

* Clarify error message for min_samples_split. (#8167)

* Upgrade html documentation to jQuery v3.1.1 (#8145)

* removed stray space in '__main__ ' (#8203)

* DOC additional fixes to 20 newsgroups to prevent TypeError (#8204)

* removed stray space in '__main__ '

* fixed pipeline bug causing TypeError and removed incorrect keyword argument

pipeline was not created properly (both classes had .predict) causing
TypeError to be thrown.

Also removed incorrect keyword argument loss=“l2” for LinearSVC

* DOC add missing parentheses in TfidfTrasnformer docstring

* TRAVIS fix flake8_diff.sh check_files (#8208)

when $files is empty.

A sphinx-gallery updated introduced flake8 violations which were not
ignored because $files was empty so all the files were checked instead
of only the examples.

* [MRG+1] Fixes #8198 - error in datasets.make_moons (#8199)

* [MRG + 2] [MAINT] Update to Sphinx-Gallery 0.1.7 (#7986)

and use latest version of sphinx in CircleCI

* [MRG+1] Add prominent mention of Laplacian Eigenmaps (#8155)

* Add prominent mention of Laplacian Eigenmaps being the algorithm that Spectral Embedding implements

* MNT/BLD Use GitHub's merge refs to test PRs on CircleCI (#8211)

Unlike other CIs, CircleCI does not test the merge commit of a PR with
its base branch. Instead it tests the PR's head commit. The problem with
this is the PR's status could differ when it is merged with the base
branch. For instance, if one needs changes in the base branch to get
their PR to pass, they must rebase/merge to get them into the history
the PR. This normally isn't a problem, but sometimes things do go wrong
when doing this merge/rebase, which adds a new unneeded difficulty.
Alternatively, a passing PR could turn out to fail when merged into the
base branch because some new content in the base branch was not tested
against in the PR.

To solve these issues, we checkout the merge ref for a PR (if it is a
PR) from GitHub. However, it should be noted that the merge ref can be
out-of-date in some cases w.r.t. the base branch. Still this is the
commonly used strategy on Travis CI and AppVeyor. If we had enough info,
we could ideally terminate a build that has merge conflicts.
Unfortunately it doesn't seem that CircleCI gives us this info.

* FIX Ensure coef_ is an ndarray when fitting LassoLars (#8160)

* Fix gh-1615: ensure self.coef_ is an ndarray

* [MRG+3] FIX Memory leak in MAE; Use safe_realloc; Acquire GIL only when raising; Propagate all errors to python interpreter level (#7811) (#8002)

* FIX MAE reg. criterion: Use safe_realloc to avoid memory leak

* Release GIL in safe_realloc and clean up scaffolding

* As gil is released in safe_realloc, no need of a with gil block

* Use except * to propagate error in all cdef functions

* Don't use except * for functions that return python objects

* Don't use except * for the comparison function passed to qsort

* Omissions and Errors

* Use safe_realloc now that gil is released there

* Fix realloc size

* Acquire GIL only if we need to raise

* Use except * more judiciously; Release gil only when raising; Add comments to clarify

* Actually that was unneeded; realloc will also allocate for the first time

* StackRecord*, PriorityHeapRecord* to fused type realloc_ptr; Use safe_realloc

* Use except -1 to propagate exceptions. This should avoid overheads

* Fix docstrings and add return 0 to reset methods

* TYPO

* REVIEW Remove redundant MemoryError raising calls

* Call sorted on lfw folder path contents (#7648)

when fetching so results are consistent across operating systems

* FIX Issue #8173 - pass n_neighbors in MI computation (#8181)

* TST/FIX Add check for estimator: parameters not modified by `fit` (#7846)

  ensure that estimators only add private attributes and attributes with
  trailing _

  in cases when existing estimators don't follow this new rule, we deprecate the
  attributes and make them follow this rule

* [MRG] #8218: in FAQ, link deep learning question to GPU question (#8220)

* linked deep learning question to gpu question.

* fix wording.

* CI remove obsolete comment

* ENH warn in classification_report when target_names doesn't equal labels size (#7802)

* Added warning for classification_report when target_names doesn't equal labels size and tests for such a case.

* [MRG] Fix aesthetic example roc crossval (#8232)

* Fix esthetic example roc crossval

* Test sphinx extensions doctests only on Circle. (#8228)

Some sphinx extensions import sphinx in their doctests.

* TST Change rstrip() to truncation in test function (#8237)

* DOC Fixing a bug where entropy included labeled items (#8150)

* Incorrect number of samples in One Hot Encoder example (#8255)

* [MRG] make the ransac example slightly more terse, improve range of plot (#8258)

* Cosmetic changes to rigde path example (#8260)

Change the name of ridge object clf to ridge and also
reinstantiate ridge in every iteration.

Issue: #8256

* DOC structure for related projects (#8257)

Also adds xgboost and eli5

* docs: related_projects.rst: fixes xgboost link (#8270)

* MAINT add Python 3.6 classifier in setup.py

* TST: added test that sample_weight can be a list (#8261)

* [MRG] Remove DeprecationWarnings in examples due to using floats instead of ints (#8040)

* [MRG] loss function plot y-label slightly confusing (#8283)

* Correct y label in plot

* Remove Space

* DOC more explicit guidelines for WIP (#8299)

* [MRG+1] Fix bench_rcv1_logreg_convergence.py by adding get_max_squared_sum

* [WIP] Fixes #8289 added get_max_squared_sum

* Fixes #8289 made recommended changes

* [MRG+1] Refactor birch-documentation (#8298)

* Refactor birch-documentation

* move

* class tags

* pep8

* [MRG] Diabetes example with GridSearchCV (#8268)

* DOC add missing release date

* [MRG+1] Enable codecov for coverage report (#8311)

* Replace coveralls from codecov for coverage report

Replacing code coverage tool from coveralls to codecov.

Issue: #8305

* Add codecov badge for coverage report

Replace coveralls badge from codecov badge

* Added Zopa testimonial (#8309)

* DOC: Remove superfluous assignment in tutorial. issue #8285 (#8314)

* remove assignment

per recommendation

* Fix doctests

* [MRG+1] Remove the MLComp text categorization example (#8264)

and deprecate load_mlcomp.

* FIX Add a missing space to an exception message in resample function (#8320)

* [MRG+1] Accept keyword parameters to hyperparameter search fit methods (#8278)

* ENH Accept keyword parameters to hyperparameter search fit methods

Deprecate ``fit_params`` as a constructor argument to the hyperparameter search classes and instead accept keyword parameters to the ``fit`` methods. This makes the ``fit`` methods of these functions conform to the Estimator API and allows the use of hyperparameter search functions in other CV utility functions such as ``cross_val_predict``.

* CR: Expanded tests, remove deprecated use in Ridge

* Make tests consistent in Python 2 and 3

* [MRG+1] Add classes_ parameter to hyperparameter CV classes (#8295)

* Add sample_weight parameter to cohen_kappa_score (#8335)

* Remove redefinition of k_fold in model_selection.rst (#8330)

* spelling mistake (#8341)

* DOC Updated documentation for scoring parameter (#8346)

Updated documentation for scoring parameter of LogisticRegressionCV to make it consistent with GridSearchCV documentation. Fixes #8333

* [MRG+2] ENH: used SelectorMixin in BaseRandomizedLinearModel (#8263)

* ENH: used SelectorMixin in BaseRandomizedLinearModel

* FIX: added get_support to return _get_support_mask

* FIX: added docstring for get_support()

* DOC: added bug fix entry to whats_new

* FIX: removed redundant get_support()

* [MRG+3] ENH Caching Pipeline by memoizing transformer (#7990)

* ENH Caching Pipeline by memoizing transformer

* Fix lesteve changes

* Fix comments

* Fix doc

* Fix jnothman comments

* DOC: added explanation for LARS (#8310)

* DOC add example regarding feature scaling (#7912)

also add load_wine to datasets

* [MRG+1] Fix description of l1_ratio for MultiTaskElasticNet (#8343)

* Fix l1_ratio documentation of MultiTaskElasticNet.

* Update l1_ratio doc of MultiTaskElasticNetCV

* Fix tests on numpy master (#8355)

numpy.apply_along_axis has changed behaviour when the function passed
in returns a 2d array

* Change "observations" to "features" in description of LassoLarsCV (#8362)

The description of LassoLarsCV compared the number of samples with the number of observations, but it was meant to compare the number of samples to the number of features (or dimensions) of the data. I changed "observations" to "features" in the following sentence:
> However, :class:`LassoLarsCV` has the advantage of exploring more relevant values of `alpha` parameter, and
if the number of samples is very small compared to the number of observations, it is often faster than :class:`LassoCV`.

* TRAVIS revert flake8 version to 2.5.1

* DOC add missing bugfix to what's new

* FIX/MAINT: update my mail etc (#8375)

* update mail etc

* fix utf8

* [MRG+1] Fix ug in BaseSearchCV.inverse_transform (#8348)

* [MRG+1]  add docs that C can receive array in RandomizedLogisticRegression (#6537)

* doc: state that parameter C can receive an array

* add more details in doc, and check the dim of C

* doc: state that parameter C can receive an array

* add more details in doc, and check the dim of C

* a little modification

* minor modifications and make line length less than 79 characters

* remove the backslash and correct typos

* meet PEP8's E128 requirement

* use .format and add a test

* fix typo (#8390)

* DOC updated IRC url to working one (#8383)

* Explain the meaning of X_m in modules/tree doc. (#8398)

* [MRG] Add the meaning of MRG and MRG+1 in the PR in docs. (#8406)

* [MRG] Make tests runnable with pytest without error (#8246)

* Make tests runnable with pytest without error.

Errors were due to pytest quirks with (deprecated) yield support.

* Add pytest build on Travis

and tweak pytest settings in setup.cfg

* Tweak comment

* plot iso-f1 curves in plot_precision_recall (#8378)

* Ignore py.test generated .cache folder

* [MRG+1] FIX AdaBoost ZeroDivisionError in proba #7501 (#8371)

* FIX AdaBoost ZeroDivisionError in proba #7501

* FIX AdaBoost ZeroDivisionError in proba #7501 - tests corrected

* FIX AdaBoost ZeroDivisionError in proba #7501 - tests corrected

* FIX #7501 improvements suggested by lesteve introduced

* FIX #7501 whats_new file updated

* Tweak in rst

* [MRG+1] Fix pickling bug  due to multiple inheritance & __getstate__ (#8324)

Fixes #8316


* Don't use test classes to group tests

* only use formatting for parts of the string that change

* Flake 8 column limit

* Make the modification of the estimator more explicit in the tests

* As suggested in code review, prefer formatting over two literals

* Also assert, that __setstate__ overwriting works in mixin

* Remove cache property

* Use assertion functions from sklearn.utils.testing

* remove the protocol argument in tests

* Rename attributes to better convey their purpose

* Revert change of module in TreeNoVersion

* Adhere to column-limit

* changelog entry

* Fix commit message

* [MRG+1] Fix message formatting in exception (#8319)

when the missing class label is a string.

* DOC Modify plot_gpc_iris.py for matplotlib v2 (#8385)

Add edgecolors option in scatter plot

Issue: #8364

* DOC svm kernel functions docs: rbf equation fixed (#8356) (#8420)

* [MRG+2] Fixed assumption fit attribute means object is estimator. (#8418)

* [MRG] FIX lasso/elasticnet example did not add noise to simulated data. (#8427)

The first argument of np.random.normal is the mean of the distribution, and
not the output shape. The example thus did not add noise but only an intercept
to the model.

* Travis add coverage to Python 3 build and oldest version build (#8435)

* [MRG] Remove unnecessary backticks around parameter name in docstrings (#8432)

* [MRG+1] Refactoring plot_iris svm example. (#8279)

* [MRG] Fix Parameters in tutorials (#8345)

* [MRG+1] Fixes incorrect output when input is precomputed sparse matrix in DBSCAN. (#8339)

* DOC fix MultiTaskElasticNet doc (#8442)

* Travis: tweak test_script.sh (#8444)

* --with-coverage was repeated twice and nose-timer options were disabled
* make test-doc uses nosetests so skip if if USE_PYTEST is true

* [MRG+1] Add note about the size of default random forest model #6276 (#8437)

* [MRG] Add MAE formula in the regression criteria docs. (#8402)

* DOC describe scikit-learn-contrib in related projects and contributing docs (#8440)

* DOC Fix default value in RandomizedLasso (#8455)

* [MRG+1] FIX/DOC Improve documentation regarding non-determinitic tree behaviour (#8452)

* FIX/DOC Improve documentation regarding non-determinitic tree behaviour

* FIX correct max_features

* Correct default value of reg_covar in gaussian_mixture. (#8462)

* Save predictions in diabetes_y_pred (#8241)

- No need for `regr.predict(diabetes_X_test)` to run multiple times.
- Use `sklearn.metrics.mean_squared_error`.
- Use `sklearn.metrics.r2_score`, instead of `regr.score`, which
  runs `regr.predict` again.

* DOCS add scikit-plot to Related Projects (#8421)

* Fix doc Multitasks (#8473)

* Correct figure number + matplotlib 2 (#8483)

* DOC example of extracting true positive, false negative, etc. (#8469)

* DOC correct typo in kneighbors parameter documentation. (#8495)

* [MRG+1] BUG: fix svd_solver validation in PCA.fit (#8496)

* BUG: fix svd_solver validation in PCA.fit

* TST: add test of pca svd_solver

* Added check_X_y to lasso_stability_path() (#7534)

* codecov: disable comments (#8502)

* [MRG] Modify Svm examples for matplotlibv2 comp (#8456)

* Modify plot_custom_kernel for matplotlib v2 comp

Add `edgecolors` attribute in scatter plot for better visualization
in matplotlib version 2

Issue: #8364

* Modify plot_oneclass.py for matplotlib v2 comp

Add `edgecolors` attribute to scatter plot for better
visualization in matplotlib version 2

Issue: #8364

* Modify plot_rbf_parameters for matplotlib v2

Add `edgecolors` attribute to scatter plot for
better visualization.

Issue: #8364

* Modify plot_separating_hyperplane_unbalanced for matplotlib v2

Add `edgecolors` attribute to scatter plot for better visualization.

Issue: #8364

* Modify plo_svm_kernels for matplotlib v2

Add `edgecolors` attribute to scatter plot for better
visualization.

Issue: #8364

* Modify plot_svm_margin for matplotlib v2 comp

Add `edgecolors` attribute to scatter plot for better
visualization.

Issue: #8364

* Modify plot_svm_nonlinear for matplotlib v2

Add `edgecolors` attribute to scatter plot for matplotlib
version 2 compatibility

Issue: #8364

* Modify file for remove flake8 error

Remove extra white space.

Issue: #8364

* turn comments off in codecov

* add html-noplot and changed help message to make.bat (#8524)

* add html-noplot and help message to make.bat

* changed spaces to tab in make.bat help

* changed all spaces to tabs in make.bat update

* [MRG+1] Repeated K-Fold and Repeated Stratified K-Fold (#8120)

* Add _RepeatedSplits and RepeatedKFold class

* Add RepeatedStratifiedKFold and doc for repeated cvs

* Change default value of n_repeats

* Change input parameters of repeated cv constructor to n_splits, n_repeats, random_state

* Generate random states in split function rather than store it beforehand

* Doc changes, inheriting RepeatedKFold, RepeatedStratifiedKFold from _RepeatedSplits and other review changes

* Remove blank line, put testcases for deterministic split in loop and add StopIteration check in testcase

* Using rng directly as random_state param to create cv instance and added a check for cvargs

* Fix pep8 warnings

* Changing default values for n_splits and n_repeats and add entry in changelog

* Adding name to the feature

* Missing space

* modify disadvantage (#8521)

[MRG+2] modify disadvantage

* fix deprecated comparison to string in GP (#8518)

* [MRG+2] referred reliability diagrams and added citations (#8527)

* [MRG] Separated regression metrics from other metrics in test_sample_weight_invariance in metrics/tests/test_common.py  (#8537)

* Separated tests for regression features in test_sample_weight_invariance

* Fixed pep8

* Removed unecessary check for regression

* Updated regression metrics

* Joel's suggestions

* [MRG] removed download_url from setup.py (#8513)

* [MRG+1] Fixes #7578 added check_decision_proba_consistency in estimator_checks (#8253)

* [MRG+2] addresses #8509 improvements to f_regression documentation (#8548)

* clarify role of the function and streamline introduction

* added feature selection methods to see also

* completed see also

* fixed pep related formatting for flake8checks.

* fixed extra whitespace flake8 problems, remaining failure is a copied see all line from another function, the line is over by a period, does not make sense to newline that.

* one more whitespace

* FIX small pep8 error.

* [MRG] Update joblib to 0.11 (#8492)

Use pip rather than easy_install in copy_joblib.sh. Also need to remove joblib/testing.py to avoid pytest dependency.

* [MRG] DOC More detailed pull request and fork instructions (#8530) (#8538)

* [MRG+2] use manylinux dev wheels for numpy / scipy (#8536)

* MAINT: use manylinux dev wheels for numpy / scipy

Use daily manylinux wheels for numpy and scipy, instead of
soon-to-be-discontinued per-commit Precise wheels.

* BF: add back ATLAS install for ubuntu build entry

scikit-learn can link against BLAS libraries still at the same location
as they were duing numpy build.

* [MRG + 1] Fix gradient boosting overflow and various other float comparison on == (#7970)

* reintroduced isclose() and flake8 fixes to fixes.py

* changed == 0.0 to isclose(...)

* example changes

* changed back to abs() < epsilon

* flake8 convention on file

* reverted flake8 fixes

* reverted flake8 fixes (2)

* np.finfo(np.float32).tiny instead of hard coded epsilon 1e-150

* reverted to 1e-150

* whats new modified

* [MRG+1] DOCATHON : Provide more intuition on perplexity in the documentation,… (#8551)

* DOC fix typo (#8560)

* Remove unneeded ** in CONTRIBUTING.md

* DOC better link in CONTRIBUTING.md (#8561)

* FIX/TST fix bug variable name for pure test (#8562)

* [MRG+1] add edgecolor to plot_pca_iris.py (#8514)

for better rendering with matplotlib 2

* [MRG] remove "matplotlib <1" compatibility code from manifold  examples  (#8552)

* [MRG + 1] Correct typo in cross decomposition example (Fixes #8307) (#8578)

* changed plsca to cca

* corrected variable plsca in line 56-57

* [MRG+1] Added v2 Custom Search API and fixed css placement for search box (#8542)

* Added v2 Custom Search API and fixed css placement for search box

* Tweaked CSS and hid the branding section

* Button made cute and override some css to make it look good for old api and new api alike

* Better button placement

* Colored the search results and pagination section

* [MRG+1] Improved docstring for permutation_test_score (#8379 and #8564) (#8569)

* Correct formulas in Doc of Gaussian Process Kernels (#8571)

Correct the Rational quadratic kernel and ExpSineSquare
kernel formulas.

Issue: #8553

* [MRG] Modify Classification module example matplotlibv2 (#8516)

* FIX s/algorithm/solver/ in bench_mnist.py

* [MRG + 2] EHN additional test for trees regarding fitting behaviour with constant features (#8580)

* TST add test checking the behaviour of constant/no-constant features

* FIX/TST factorize test

* TST Add additional constant features

* FIX/TST remove ExtraTree from test

* [MRG+1] FIX Correct depth formula in iforest  (#8576)

* Fixed depth formula in iforest

* Added non-regression test for issue #8549

* reverted some whitespace changes

* Made changes to what's new and whitespace changes

* Update whats_new.rst

* Update whats_new.rst

* fixed faulty whitespace

* faulty whitespace fix and change to whats new

* added constants to iforest average_path_length and the according non regression test

* COSMIT

* Update whats_new.rst

* Corrected IsolationForest average path formula and added integer array equiv test

* changed line to under 80 char

* Update whats_new.rst

* Update whats_new.rst

* reran tests

* redefine np.euler_gamma

* added import statement for euler_gammma in iforest and test_iforest

* changed np.euler_gamma to euler_gamma

* fix small formatting issue

* fix small formatting issue

* modified average_path_length tests

* formatting fix + removed redundant tests

* fix import error

* retry remote server error

* retry remote server error

* retry remote server error

* re-added some iforest tests

* re-added some iforest tests

* [MRG + 1] Return correct ridge parameter alpha_ and lambda_ for Bayesian ridge regression (#8567)

* Return correct ridge parameter alpha_ and lambda_ for regression

* Add test for coefficients and fix style

* Move sklearn.utils.testing to a more reasonable position.

* Make flake8 happy

* Code cleanup and entry in whats_new.rst

* Fix typo in documentation (#8600)

x_i should be in R^m (as n is the number of features)

* [MRG+1] GaussianProcessRegressor: faster prediction of std (#8591)

* Improve warning message in ElasticNet (#8565)