In this assignment, you are to write a C++ program (using Visual C++ 2015) that reads training data in WEKA arff format and generates ID3 decision tree in a format similar to that of the tree generated by Weka ID3. Please note the following:

Your algorithm will use the entire data set to generate the tree. You may assume that the attributes (a) are of nominal type (i.e., no numeric data), and (b) have no missing values.

In general, the basic ID3 algorithm uses entropy measure to select the best attribute to divide the data set. It continues to select attribute for further branching (based on the subset of data belong to that branch) until either (a) all attributes have been used, or (b) all instances under a node belong to the same class. This ensures a 0% error rate on the training set although it may not work the best with future data due to over-fitting.