Philip Hyunsu Cho
3c72654e3b
Revert "Fix #3485 , #3540 : Don't use dropout for predicting test sets" ( #3563 )
...
* Revert "Fix #3485 , #3540 : Don't use dropout for predicting test sets (#3556 )"
This reverts commit 44811f233071c5805d70c287abd22b155b732727.
* Document behavior of predict() for DART booster
* Add notice to parameter.rst
2018-08-08 09:48:55 -07:00
Philip Hyunsu Cho
44811f2330
Fix #3485 , #3540 : Don't use dropout for predicting test sets ( #3556 )
...
* Fix #3485 , #3540 : Don't use dropout for predicting test sets
Dropout (for DART) should only be used at training time.
* Add regression test
2018-08-05 10:17:21 -07:00
Laurae
5c13aa0a8a
GLM test unit: make run deterministic ( #2147 )
2017-03-24 08:54:39 -07:00
Yuan (Terry) Tang
090b37e85d
Bumped up err assert in glm test ( #1792 )
2016-11-20 18:23:19 -06:00
Yoshinori Nakano
949d1e3027
add Dart booster ( #1220 )
2016-06-08 14:04:01 -07:00
Vadim Khotilovich
9a48a40cf1
Fixes for multiple and default metric ( #1239 )
...
* fix multiple evaluation metrics
* create DefaultEvalMetric only when really necessary
* py test for #1239
* make travis happy
2016-06-04 22:17:35 -07:00
Faron
ad3f49e881
[py] eta decay bugfix
2016-04-30 15:51:57 +02:00
sinhrks
6bab164d80
Bug mixing DMatrix's with and without feature names
2016-04-30 14:42:57 +09:00
sinhrks
8fc2456c87
Enable flake8
2016-04-24 17:32:31 +09:00
terrytangyuan
803a6fe474
Separate dependencies and lightweight test env for Python
2016-02-28 20:11:10 -06:00
