GitLab

[tests][python] added tests for metrics' behavior and fixed case for multiclass task with custom objective (#1954)

* added metrics test for standard interface

* simplified code

* less trees

* less trees

* use dummy custom objective and metric

* added tests for multiclass metrics aliases

* fixed bug in case of custom obj and num_class > 1

* added metric test for sklearn wrapper

* added get_data method

* hotfix

* added warning for other data types

* reworked according to review comments

* minor addition to FAQ

* added test

* [python] fix creating train_set in fit

https://github.com/Microsoft/LightGBM/blob/cc99f0d36ae929eb02b22a072823ab7c6d3155ab/python-package/lightgbm/sklearn.py#L519
may False even if valid_data[0] is X and valid_data[1] is y actually, because `check_X_y` might return copy of X and y.
https://scikit-learn.org/0.20/modules/generated/sklearn.utils.check_X_y.html

cf. https://github.com/Microsoft/LightGBM/pull/451

* use assertIn

* fix RF's bugs

* fix tests

* rollback num_iterations

* fix a bug and reduce memory costs

* reduce memory cost

* add pandas_categorical to returned json

* save pandas_categorical to model file

* added regression test

* removed excess conversion to list

* removed deprecated line

* hotfix

* break huge lines in sklearn tests

* break huge line in plotting tests

* break huge lines in basic tests

* multiple enhancements in engine tests

* multiple enhancements in sklearn tests

* hotfixes

* break huge lines and use with statement in C API test

* make NDCG test more strict

* fix ndcg consistency.

* more stable sorts

* Update gbdt_model_text.cpp

* Update dataset.cpp

* Update gbdt_model_text.cpp

* average predictions for constant features

* fix possible numerical issues in std::log.

* fix pylint

* fix bugs in c_api

* fix styles

* clean code for multi class

* rewrite test

* fix pylint

* skip test_constant_features

* refine test

* fix tests

* fix tests

* update FAQ

* fix test

* Update FAQ.rst

* fixed FutureWarning about cv default value

* fixed according to new check_estimator API

* fixed joblib warning

* added sklearn splitter classes in cv function

* added tests

* fix RF's bug withMAPE

* simplify rf's code & support multi-class rf

* fix bug & add test

* add more tests

* Update test_engine.py

* Update test_engine.py

* added test for pandas label of Dataset

* fix when label type is pandas DataFrame; document possible pandas Series type

* added NumberOfTotalModel and NumModelPerIteration to C_API and python-package

* fixed tests

* added tests for current_iteration, num_trees, num_model_per_iteration methods

* break huge line in test

* hotfix

* add support of refit-decay

* add refit into c_api

* add test

* update document

* Update basic.py

* Update test_engine.py

* Update basic.py

* Update test_engine.py

* fix comments

* update test

* fix the comments

* Update test_engine.py

* added new aliases for params

* run helper/parameter_generator.py

* removed useless test

* ignore vim temporary files

* add importance_type arg to sklearn API

* update documentation info

* remote a trailing space

* remove trailing space (again :))

* add instructions on importance choices to sklearn API

* drop mention of constructor in the feature type setting

* adding a test for different feture types

* remove trailing spaces, make shorter assert in feature importance type handling test

* fixing style issue introduced with the new test

* removed excess import

* added tests for plotting trees in Python

* refined module_INSTALLED mechanism

* added note about that create_tree_digraph is better than plot_tree

* fixed grammar

* fixed params description in graph plotting functions

* clarified types of attributes in their descriptions

* increased readability of graphs by adding spaces

* added precision parameter to plot tree functions

* fixed docs

* reworker predict method of sklearn wrapper

* fixed encapsulation

* added test

* fixed consistency between docstring and params docs

* fixed verbose

* replaced predict_proba with predict in test

* fixed verbose again

* fixed fraction params descriptions

* added description of skip_drop and drop_rate constraints

* fixed subsample_freq consistency with C++ default value

* fixed nice look of params list

* made force splits json file example clickable

* fixed nice look of metrics list and added comma

* reduced warning in test about same param specified twice

* replaced pred_parameter with **kwargs in predict method

* added test for **kwargs in predict method

* fixed warnings

* fixed pylint

* updated pep8 to pycodestyle

* fixed E722 do not use bare 'except'

* fixed W605 invalid escape sequence '\*'

* fixed W504 line break after binary operator

* ignore W605 invalid escape sequence '\*' in nuget builder

* made pycodestyle happy

* Read and write datsets from hdfs.
* Only enabled when cmake is run with -DUSE_HDFS:BOOL=TRUE
* Introduces VirtualFile(Reader|Writer) to asbtract VFS differences

* fix early stopping edge case

* fix message.

* fix tests

* fix GPU tests.

* remove protobuf

* add version number

* remove pmml script

* use float for split gain

* fix warnings

* refine the read model logic of gbdt

* fix compile error

* improve decode speed

* fix some bugs

* fix double accuracy problem

* fix bug

* multi-thread save model

* speed up save model to string

* parallel save/load model

* fix some warnings.

* fix warnings.

* fix a bug

* remove debug output

* fix doc

* fix max_bin warning in tests.

* fix max_bin warning

* fix pylint

* clean code for stringToArray

* clean code for TToString

* remove max_bin

* replace "class" with typename

* add init_score & test cpp and python result consistency

* try fix common.h

* Fix tests (#3)

* update atof

* fix bug

* fix tests.

* fix bug

* fix dtypes

* fix categorical feature override

* fix protobuf on vs build (#1004)

* [optional] support protobuf

* fix windows/LightGBM.vcxproj

* add doc

* fix doc

* fix vs support (#2)

* fix vs support

* fix cmake

* fix #1012

* [python] add network config api  (#1019)

* add network

* update doc

* add float tolerance in bin finder.

* fix a bug

* update tests

* add double torelance on tree model

* fix tests

* simplify the double comparison

* fix lightsvm zero base

* move double tolerance to the bin finder.

* fix pylint

* clean test.sh

* add sklearn test

* remove underline

* clean codes

* set random_state=None

* add last line

* fix doc

* rename file

* try fix test

* Update test_engine.py

* Update test_engine.py

commit c9e123f24fcbb159c04e6694c7f830530bb2f27e
Author: Guolin Ke <i@yumumu.me>
Date:   Wed Oct 18 10:00:19 2017 +0800

    change default max_cat_to_onehot

commit 805a5c3125b9979d634922e1708877fa0fec80c6
Author: Guolin Ke <i@yumumu.me>
Date:   Tue Oct 17 22:57:18 2017 +0800

    use one hot coding for the small cats

* many fixes for categorical feature

* add l2 to categorcial split.

* remove useless file

* update version

* add cat_l2

* update appveyor verison

* remove file

* fix tests.

* change default cat_l2 value

* fix a bug in bin finder

* change default cat_smooth_ratio

* refine categorical split

* a bug fix

* fix a bug

* refine categorical split

* add test

* added test for sklearn handle categorical features

* use raw X, y in sklearn wrapper in case of pandas.DataFrame

* fixed probs

* Fix feature attributions for regression models and add Python bindings

* Address pylint issue

* Lazy fix missing tree depth info

* disabled logs from compilers; fixed #874

* fixed safe clear_fplder

* added windows folder to manifest.in

* added windows folder to build

* added library path

* added compilation with MSBuild from .sln-file

* fixed unknown PlatformToolset returns exitcode 0

* hotfix

* updated Readme

* removed return

* added installation with mingw test to appveyor

* let's test appveyor with both VS 2015 and VS 2017; but MinGW isn't installed on VS 2017 image

* fixed built-in name 'file'

* simplified appveyor

* removed excess data_files

* fixed unreadable paths

* separated exceptions for cmake and mingw

* refactored silent_call

* don't create artifacts with VS 2015 and mingw

* be more precise with python versioning in Travis

* removed unnecessary if statement

* added classifiers for PyPI and python versions badge

* changed python version in travis

* added support of scikit-learn 0.18.x

* added more python versions to Travis

* added more python versions to Appveyor

* reduced number of tests in Travis

* Travis trick is not needed anymore

* attempt to fix according to https://github.com/Microsoft/LightGBM/pull/880#discussion_r137438856

* fixed sklearn test on python 2.7

* commit to show that problem has been solved

* come back to python 3.6

* removed warnings check

* improved sklearn interface; added sklearns' tests

* moved best_score into the if statement

* improved docstrings; simplified LGBMCheckConsistentLength

* fixed typo

* pylint

* updated example

* fixed Ranker interface

* added missed boosting_type

* fixed more comfortable autocomplete without unused objects

* removed check for None of eval_at

* fixed according to review

* fixed typo

* added description of fit return type

* dictionary->dict for short

* markdown cleanup