GitLab

* speed up the AUC metric by parallel sort.

* fix unix compile.

* fix template in unix.

* remove warning.

The result of a right-shift of a signed negative number is
implementation-dependent.
cf. https://msdn.microsoft.com/en-us//library/336xbhcz.aspx

Fix #519
Fix #521

* refine prediction logic.

* fix test.

* fix out_len in training score of Dart.

* improve predict speed for high dimension data.

* many refactors.

* remove multi_loglossova.

* fix tests.

* avoid using lambda function.

* fix some format.

* reduce branching.

* Added possibility to compile without openmp (useful for clang on Mac)

* Added CMake option to create dylib on MacOS

* src & callbacks

* lgb.Booster and utils

* cv

* wip lgb.Dataset

* lgb.Dataset

* lgb.Predictor

* lgb.train

* typos

* add flags to facilitate macosx compilation

* fix basic_string template error with clang

* most unfortunate mode of development

* fixup tests

* last test

* roxygen

* roxygen v5.x.x

* fix typo

* fix hessians to approximate hessians with Gaussian function

* fix ApproximateHessianWithGaussian

* take fabs of gradient

* use atan(1) to calculate pi

* fix pi

* Add L1 objective function

* fix hessians

* update

Categorical feature support (#108)

1. merge python-package
2. add dump model to json
3. fix bugs
4. clean code with pylint
5. update python examples