Decision Tree Induction Using Evolutionary Algorithms: ASurvey

Abstract

An evolutionary methods for an induction-based decision trees made a wide step development in machine learning field.In this context, the majority of researches recently concentrates on techniques that use developing decision trees as an alternative to the traditional heuristic top-down divide-and-conquer strategy.Evolutionary algorithms play an important role in improving decision tree classifier parts.The main contributions of our article are twofold, first it provides a survey of evolutionary algorithms with decision trees.Second, it reviews a taxonomy that encompasses techniques mentioned above as a backbone in creating enhanced decision trees, and an evolved construction components of decision trees.The article covering researches in the period 2011-2023, these researches proposed different evolution paradigms encompasses :feature categorization, splitting nodes , complex to simple decision rules, tree size, etc parameters of DT.Finally, a detailed scenarios and results had been analyzed, highlighting the weaknesses and areas of strength with respect to processing time, accuracy, and required space.