Problem 1: CART
Given dataset below you want to build a regression tree, where x1 and x2 are the features and y is the label.
x1 x2 y 2 8 positive 3 4 negative 12 6 positive 2 3 negative 8 -4 positive 6 3 positive
(a) Which feature(xl or x2) would you use as your first split on your decision tree, why?
(b) Draw the regression tree built by the ID3-Algorithm introduced in class
(c) If we add a new categorical feature x3 as seen below, how would you account the new change? (Hint: think about what it means for the algorithm to "split" on a categorical feature.)
x1 x2 x3 y -2 8 B positive 3 4 R negative 12 6 G positive 2 3 R negative 8 -4 B positive 6 3 G positive