xgb-subsampling-param-desc.md

col_sample_by parameters "stack" on top of each other if several are specified. That is, given a set of 24 columns, for col_sample_by_tree=0.5 andcol_sample_by_node=0.5,col_sample_by_tree samples 12 columns, reducing the available, unsampled column pool to 12. col_sample_by_node then samples half of the remaining pool, so each node samples 6 columns.

This algorithm will always sample at least one column.