How to tune tree parameters in GBM in R?

r
gbm

#1

Hello people,

I was using gbm library to perform gradient boosting in R. There are several parameters that can be tuned in Python according to this blog on Analytics Vidhya.

However, when I saw the documentation of gbm function in R, I didn’t find any such parameters. See here
Is there no provision for setting pruning parameters like ‘min-split’ in R? Please suggest some method for doing this.

Thanks in advance!

Regards,
Shashwat


#2

The implementation of GBM algorithm in R and python is different, and hence the difference. I would recommend you to see this video to understand GBM in R


#3

Thanks @jalFaizy,

There are relatively fewer parameters in R for GBM. We can set the important ones however like interaction.depth and minobsinnode. These do prune the trees formed quite effectively.


#4

You can check out the GBM function under the library h2o
Thanks,
Tapo