mtry maps to wrong parameter for XGBoost

[smingerson]

From ?boost_tree, we have

mtry: The number of predictors that will be randomly sampled at each split when creating the tree models.

In the model argument translation for XGBoost, we have mtry maps to colsample_bytree.

However, the XGBoost documentation says

colsample_bytree is the subsample ratio of columns when constructing each tree. Subsampling occurs once for every tree constructed.

I don't read this and mtry as meaning the same thing. In fact, mtry sounds more like this XGBoost documentation excerpt:

colsample_bynode is the subsample ratio of columns for each node (split). Subsampling occurs once every time a new split is evaluated. Columns are subsampled from the set of columns chosen for the current level.

[juliasilge]

Hmmmm, I think you are right. The other boosted tree implementation that we support with mtry is from sparklyr, where it is called feature_subset_strategy and says:

The number of features to consider for splits at each tree node. See details for options.

So also clearly about splitting at nodes.

[github-actions]

This issue has been automatically locked. If you believe you have found a related problem, please file a new issue (with a reprex: https://reprex.tidyverse.org) and link to this issue.