Uses OOB forest weights to predict both xvar and yvar, then computes the mean normalized MSE over all predictor and response coordinates. Lower is better; when predictor and response dimensions differ, every coordinate still receives equal weight.
Uses OOB forest weights to predict both xvar and yvar, then computes the mean normalized MSE over all predictor and response coordinates. Lower is better; when predictor and response dimensions differ, every coordinate still receives equal weight.