GitLab

* expose feature importance to c_api

* support type=gain

* remove dump model from examples and tests temporarily because it's unstable

* use double instead of float

* Reorganize documentation references

* Fix R-package link name

* updated scikit-learn interface

* fixed better description

* updated set_params()

* removed backward compatibility

* removed excess lines

* replaced pop with setdefault

* added deprecated warnings

* added tests

* fixed links while viewing from wiki page

* fixed typos

* fixed repeated word and added markdown in FAQ.md

* fixed link to VS

* try to fix python memory leak

* fix parameter error.

* free old handle in _load_model_from_string .

* fix best_score & best_iteration

* refine logics.

* fix tests.

* output warning when only set `max_depth`

* Remove check of max_depth

max_depth can take both positive and negative value

* python-package: fix version

* remove version variable

* Prepare merging parameters

* Add parameters from params for lgb.train

* Usage of params in lgb.cv for rounds/early stop

* Double bracket (1)

* Double brackets (2)

* Explain individual predictions using SHAP value feature attributions

* Address code review

* check params

* add test case

* fix pylint

* Fix Invalid decision type error while plotting tree in Python

* Fix Invalid decision type error while plotting tree in Python (modified = to ==)

* Updated link to Development Guide in README.md

* Updated link to DevelopmentGuide in docs/README.md

* Revert "[R-package] Prevent remembering parameters (#796)"

This reverts commit c795e2c8e12e8fe989bde0ab4ee59c2301325814.

* Use model load/unload trick for free-ing up memory

* Missing comma

* Hand copy manual update (1/2)

* Hand copy manual update (2/2)

* Enable simple cloning of dataset

* Add remodel logical to Dataset

* Allow remodeling of data

* Missing comma

* Use name instead of assignment for R6 class

* There should be no need to deep copy ALL objects