A Fuzzy Decision tree approach for imbalanced data classification

Abstract

Recently, new Fuzzy Decision Tree (FDT) approaches have been developed for doing classification tasks. In this paper, one of these FDTs is adapted for performing the imbalanced classification tasks. First, our proposed method utilizes k-means algorithm to cluster the majority class samples into some clusters. Then, each cluster is labeled as a new class and thereby the binary imbalanced classification problem is converted to the multi-class classification problem. Eventually, FDT algorithm is employed for classifying the new data set. The obtained results show that our proposed method outperforms almost all the other fuzzy rule based approaches over highly imbalanced data sets.