A PROC HPFOREST model is trained with VARSTOTRY=5 on a dataset with…

Question

A PROC HPFOREST model is trained with VARSTOTRY=5 on a dataset with 40 predictor variables. What does VARSTOTRY=5 control?

Accepted Answer

The number of randomly selected candidate variables evaluated at each node split during tree construction, implementing the 'feature subspace sampling' component of random forests — Here's the deal — pick the option saying VARSTOTRY=5 randomly selects 5 candidate predictors at each node split. That feature subspace sampling decorrelates trees.

Answer

The minimum number of variables required in each leaf node of every decision tree

Answer

The total number of trees built in the forest that use each variable at least once as a split variable

Answer

The maximum number of interaction terms between variables that PROC HPFOREST generates automatically

A PROC HPFOREST model is trained with VARS_TO_TRY=5 on a dataset with 40 predictor variables. What does VARS_TO_TRY=5 control?

Short Explanation

Full Explanation