The repository for learning purposes of Decision Trees in Machine Learning. Implementation of primary Decision Tree algorithms from scratch (ID3, C4.5) with multiple splitting criteria. Besides that, pruning techniques are also implemented to avoid overfitting.
The classifiers can be pruned using soft pruning technique, which involves the bacpropagation neural network algorithm to learn the rules of unpruned tree. This approach does not lose the explanatory power of the tree and also avoids overfitting.
The learned decision tree classifiers can be visualized using the graphviz library.