[R-package] categorical_features should be in lgb.Dataset

[Laurae2]

Demos to fix later:

https://github.com/Microsoft/LightGBM/blob/master/R-package/demo/categorical_features_prepare.R
https://github.com/Microsoft/LightGBM/blob/master/R-package/demo/categorical_features_rules.R

categorical_features should be used in lgb.Dataset

But also not lgb.Dataset.create.valid, otherwise:

Error in .subset2(public_bind_env, "initialize")(...) : 
  lgb.Dataset: Only can use ‘lgb.Predictor’ as predictor

Warnings of code:

[LightGBM] [Warning] Unknown parameter categorical_feature=
[LightGBM] [Warning] Unknown parameter 2,
[LightGBM] [Warning] Unknown parameter 3,
[LightGBM] [Warning] Unknown parameter 4,
[LightGBM] [Warning] Unknown parameter 5,
[LightGBM] [Warning] Unknown parameter 7,
[LightGBM] [Warning] Unknown parameter 8,
[LightGBM] [Warning] Unknown parameter 9,11,16
[LightGBM] [Warning] min_data is set=1, min_data=1 will be ignored. Current value: min_data=1
[LightGBM] [Warning] Unknown parameter categorical_feature=
[LightGBM] [Warning] Unknown parameter 2,
[LightGBM] [Warning] Unknown parameter 3,
[LightGBM] [Warning] Unknown parameter 4,
[LightGBM] [Warning] Unknown parameter 5,
[LightGBM] [Warning] Unknown parameter 7,
[LightGBM] [Warning] Unknown parameter 8,
[LightGBM] [Warning] Unknown parameter 9,11,16
[LightGBM] [Warning] Starting from the 2.1.2 version, default value for the "boost_from_average" parameter in "binary" objective is true.
This may cause significantly different results comparing to the previous versions of LightGBM.
Try to set boost_from_average=false, if your old models produce bad results
[LightGBM] [Info] Number of positive: 458, number of negative: 3542
[LightGBM] [Info] Total Bins 930
[LightGBM] [Info] Number of data: 4000, number of used features: 16
[LightGBM] [Warning] Unknown parameter categorical_feature=
[LightGBM] [Warning] Unknown parameter 2,
[LightGBM] [Warning] Unknown parameter 3,
[LightGBM] [Warning] Unknown parameter 4,
[LightGBM] [Warning] Unknown parameter 5,
[LightGBM] [Warning] Unknown parameter 7,
[LightGBM] [Warning] Unknown parameter 8,
[LightGBM] [Warning] Unknown parameter 9,11,16
[LightGBM] [Info] [binary:BoostFromScore]: pavg=0.114500 -> initscore=-2.045578
[LightGBM] [Info] Start training from score -2.045578

[guolinke]

@Laurae2 Can this be closed ?

[Laurae2]

One file seems to have still have the issue.