MAHAKIL: diversity based oversampling approach to alleviate the class imbalance issue in software defect prediction.
Abstract
∗ Kwabena E. Bennin1 , Jacky Keung1 , Passakorn Phannachitta2 , Akito Monden3 and Solomon Mensah1 1 Department of Computer Science, City University of Hong Kong, Hong Kong 2 College of Arts, Media and Technology, Chiang Mai University, Thailand 3 Graduate School of Natural Science and Technology, Okayama University, Japan {kebennin2-c,Jacky.Keung,smensah2-c}@my.cityu.edu.hk,passakorn.p@cmu.ac.th,monden@okayama-u.ac.jp ABSTRACT oversampling techniques address this concern by creating new minority defective modules to balance the minority and majority class distribution. Notwithstanding the successes attained by these techniques, they sometimes generate erroneous data instances, which lead to high false positives. They also tend to generate less diverse data points and data points very similar to existing samples because they consider only nearest neighbor samples. We propose a novel oversampling technique that generates diverse synthetic data restricted within the minority class region. This study presents MAHAKIL, a novel and efficient synthetic oversampling approach for software defect datasets that is based on the chromosomal theory of inheritance. Exploiting this theory, MAHAKIL interprets two distinct sub-classes as parents and generates a new instance that inherits different traits from each parent and contributes to the diversity within the data distribution. We extensively compare MAHAKIL with five other sampling approaches using 20 releases of defect datasets from the PROMISE repository and five prediction models. Our experiments indicate that MAHAKIL improves the prediction performance for all the models and achieves better and more significant pf values than the other oversampling approaches, based on robust statistical tests. 2 APPROACH AND UNIQUENESS Our proposed oversampling approach takes inspiration from the field of biology by considering how diverse populations of living organisms differ although they are of the same species. This phenomenon, explained by the Chromosomal Theory of Inheritance [1], justifies how an offspring inherits traits from their parents by obtaining chromosomes from each parent in equal quantity. The proposed approac