3.3 [4 pts] Which tree would you select? In other words, would you use the original decision tree or would you prune it? Briefly explain how you will decide. 3.4 [6 pts] Build the confusion matrices when using DT1 and DT2 to predict for the test set. hom 3.5 [4 pts] Compute the error rate of the DT1 and DT2 on the test set shown in the table 5. A tudha Test set 02869 Training set A B # of (+) instances # of (-) instances # of (+) instances # of (-) instances 0 0 1 22 0 1 0 1 0 1 4 3 2746 22 3 4 Table 5 7 30 6 4 6 3 6 4 1 (laf) 2 (T) 3 otoq8 2 6 (12) oltaR 3 Consider the decision trees shown in Figure 1. The decision tree in 1b is a pruned version of the original decision tree la. The training and test sets are shown in table 5. For every combination of values for attributes A and B, we have the number of instances in our dataset that have a positive or negative label. 0 B 0 B 1 bain ITC (a) Decision Tree 1 (DT1) 0 B 1 Figure 1 0 0 A B 2 (b) Decision Tree 2 (DT2)