Printer-friendly version

Completely unsupervised random forest method on Training data with ntree = 150 leads to the following error plot:

Importance of predictors are given in the following dotplot:

Accuracy improves from 50% to 67.7%.