130
120
100
95
95
90
85
85
85
85
40
 

The total for HW3 is 120: 50 for Decision Tree, 50 for Naive Bayesian, and 20 for 10-Fold Cross Validation and t-Test.

The Cars dataset turns out to be an easy one for classification - note that the dataset contains all possible combination of the attribute values. The accuracy of a correct implementation of decision tree without any pruning should be in the 90s, while for Naive Bayesian the accuracy should be in the 80s.

If you have any question about your grade, please come by my office tomorrow (Thursday, 3/10). Note that to convince me your implementation is correct, you must show that your code can work on a small dataset where the results can be manually verified.