Analysis of the prediction performance of decision tree-based algorithms

Due to their many benefits, decision trees are extensively utilized to solve a variety of classification problems in the real world. Consequently, creating efficient and effective decision trees remains a common challenge for researchers in the fields of machine learning and data mining. In the literature, various decision tree algorithms with different attribute selection criteria are introduced. Although the currently available decision tree algorithms achieve different performances, none of them can produce optimal decision trees for a range of data sets. In this study, the efficacy of two traditional decision tree techniques is contrasted with the efficacy of two more recent decision tree approaches. On eleven real-world data sets, the four methods are compared in terms of four evaluation metrics: classification accuracy, tree depth, leaf nodes, and tree construction time.