ALGORITHM 2.

Marginal VIMP index for a variable β

1:	for b = 1, …, B do
2:	Draw a bootstrap sample of the data; calculate the model OOB prediction error, Errb.
3:	Fit a second model, but without β, and calculate its OOB prediction error, ${\text{Err}}_{\beta ,b}^{\text{marg}}$
4:	end for
5:	Calculate the marginal VIMP by averaging: ${\mathit{\Delta }}_{\beta }^{\text{marg}}={\sum }_{b=1}^B[{\text{Err}}_{\beta ,b}^{\text{marg}}-{\text{Err}}_b]/B$