ID3-algorithm

Matlab implementation of the ID3 algorithm for classification: this implementation makes use of entropy and information gain to split the node of a tree. This repository contains different files:

• main.m: calls the function decision_tree_classifier(), that receives the training set, the class labels and the number of columns to consider as numerical and the test set;
• decision_tree_classifier.m: calls the function build_tree() in order to build a decision tree and then classifies the test set with the function classifier();
• build_tree.m: builds the decision tree recursively by splitting each node with the feature that maximizes the information gain.
• classifier.m: computes the classification with the tree previously built;
• H.m: function that computes the entropy of a probability vector.

Usage

Insert the training set and the corresponding labels and indicate which columns should be considered as numerical in the file main.m: it outputs the classification of the test set made by the tree built.