When I fit a regression tree model in R, I can somehow look at the plotted tree to figure out the interactions of the predictors. However, when I fit a random forest model, I cannot plot a tree because there are too many of them. I can somehow find the importance order of the predictors (using varImpPlot function), but I do not know how to detect the interactions from the model. By the way, the importance plot only gives the importance order of the predictors, how can I figure out exactly how many of them should be included for further investigation. Thank you.
-
2$\begingroup$ Addressed in the answers from Jan. 25, 2012 at stats.stackexchange.com/questions/21152/… $\endgroup$ – rolando2 Mar 4 '15 at 19:08