Paper Details

Classification is a technique used to predict categorical dependent variables. Classification is known as supervised learning since class labels are known in advance. It has applications in various areas. In this paper, we have performed a comparative study of ID3, CART, C5.0 decision tree algorithms. Decision tree algorithms are effective and easy to interpret as compared to other classification algorithms. We have used free datasets from UCI machine learning repository from different subject areas to illustrate how the recursive partitioning algorithms differ from one another. We have chosen datasets with small, medium and large number of observations to avoid any bias in the comparative analysis. We have reviewed ID3, CART, C5.0 algorithms and by experimental analysis we focused on the aspects like time taken to build the model, accuracy, depth, and breadth of the resulting classification tree, as well as size of the tree, that is, the total number of nodes.

Classification, decision tree, recursive partitioning, ID3, CART, C5.0