Hello. I assume that you are using GBR from the sklearn. See this example.

What is the difference between the data that has multiple columns and you want to use only one of them, and the data that has only one column? In both of the cases, I don't see any particular reason to use K means instead of just a simple histogram.

@oW_ I Can you give a little bit of explanation why your approach is better than mine?

Ok done. Any comments on my approach?

Yes this reduction is done for purpose I want to increase the training time. After I obtain the best parameters values I crossvalidate (using 10 folds) with all of available data.