REVIEW OF FUZZY DECISION TREES AND ITS VARIANTS

Abstract

This thesis is about the advancements that have been made in building and applying fuzzy decision trees to solve classification problems characterized by ambiguity and uncertainty. The classification with such kinds of classification methods is essentially exposed with crisp, rigid boundaries and does not capture the vagueness being transmitted from real-world data. In this regard, fuzzy decision trees provide a more flexible approach, in which such vagueness can be captured, hence enabling objects to belong to different classes with varying degrees of membership according to the principles of fuzzy logic. The thesis also discusses the variants that treat all the variables at every node, such as C fuzzy decision trees and Neuro-Fuzzy Decision Trees (N-FDTs), which bring in neural learning to improve accuracies further but still be interpretable. The concept of Intuitionistic Fuzzy Decision Trees (IFDTs) is added by incorporating the idea of hesitation parameters so they can handle uncertainty better. Overall, this work has developed frame for the induction and application of fuzzy decision trees in improving any kind of decision-making process within complex and uncertain environments.