How to speed up XGboost modelling with tidymodels

I am looking for a way to speed up XGboost performance. Using Tidymodels route with parsnip::boost_tree in regression mode. I am still fresh when it comes to XGboost. After looking up the topic on the internet, the best bet seems to use histogram-based trees and potentially use gpu for computing.


lightgbm is a lot faster. You can use that via tidymodels via the bonsai package.

You can pass histogram related options via set_engine().

Also, if you are using the tidymodels tuning functions, it is better to parallel process via tidymodels instead of the parallel processing available within each xgboost fit.

1 Like

Thanks Max! What is the most up to date tidymodels instructional? I reckon the ecosystem is developing quite steadily and some blogposts I used seem to be a bit outdated. and

1 Like

This topic was automatically closed 7 days after the last reply. New replies are no longer allowed.

If you have a query related to it or one of the replies, start a new topic and refer back with a link.