1. Given below is a summary table of a dataset. Solve the following problems by hand (and a calculator). A B C Number of instances Class = + Class = - =- T T T 5 0 F T T 0 20 T F T 20 0 F F T 0 5 T T F 0 0 F T F 25 0 T F F 0 0 F F F 0 25 Please note each row in the table represents a group of records with the same attribute values. There are 100 records in total: (5 +20 +25) positives and (20+ 5+ 25) negatives. 1) Train a DT by using the given dataset. Use entropy as the impurity measure. Grow the DT until it reaches level 3, i.e., the (deepest) leaf nodes are at level 3 (root at level 1). 2) Evaluate performance of the DT by error rate for the train set (i.e. resubstitution error). 3) Repeat the above questions, but begin with C as the split at the root this time. 4) Compare the error rates of the DTs. Which DT is better? Based on the comparison, comment on the greedy approach of the DT algorithm.