Anna Veronika Dorogush

We will also allow for one-hot-max-size > 255

You need to set thread_count for both of them to the same number, for example to 16 so that is is parallelized with same amount of threads. Plus LightGBM builds...

Actually LightGBM performs best with 1 thread per core, and it sets 32 threads by default, so the comparison needs to be changed - if we run it without hyperthreading,...

And for large datasets we have same speed (here again 16 threads, 64 leaves, no hyperthreading, and 100k docs): ``` espetrov@park:~/svn/trunk/arcadia$ python github_catboost_regr.py ('lightgbm time: ', 12.613971948623657) ('catboost time: ',...

@szilard could you try upgrading the version? This bug should have been fixed today in version 0.6.1.1. One more thing about airline data. It is a very special dataset since...

It's again a bug (the accuracy and the wait after the training), the code of the fix is already on github, but not jet on pypi. Will be there in...

The preprocessing contains data binarization and calculation of part of statistics for categorical features, loading everything on GPU. The postprocessing contains calculation of all selected statistics on categorical features, loading...

If I'm not mistaken, you should be doing issues here: https://github.com/catboost/catboost/issues I would recommend to create the new ones there, because we check that place all the time. About this...

object_importance is not yet supported for multiclass models. It would be really nice to add it though! So I'm adding help wanted label.

The initial issue is about object importance. Please create a separate issue if you have a problem with training or some other operation. As for the initial issue - object...