Description

a) On the madelon dataset, for each of k ∈ {3,10,30,100,300} train a random forest with k√trees where the split attribute at each node is chosen from a random

subset of ∼ 500 features. Use the trained trees to predict the class labels on the training and test sets, and obtain the training and test misclassification errors. Plot on the same graph the training and test errors vs number of trees k as two separate curves. Report the training and test misclassification errors in a table. (4 points)
b) Repeat point a) on the madelon dataset where the split attribute at each node is chosen from a random subset of ∼ ln(500) features. (2 points)
c) Repeat point a) on the madelon dataset where the split attribute at each node is
chosen from all 500 features. (2 points)
1