Comparison of Classification Methods on Software Defect Data Sets

Abstract

Nowadays it is difficult for us to imagine a life without devices that is controlled by software. Software quality prediction is the important process of software development processes. It is a process of utilizing software metrics such as code-level measurements and defect data to estimate software quality modules. A more useful and efficient mechanism is k Nearest Neighbor method to classify class of target data based on k nearest training dataset. By applying the concept of k-NN, we propose a new mechanism called Class Base Weighted k-NN with Biner Algorithm (CBW k-NN) to find the range of training dataset where the target data has the maximum likelihood of occurrence by Biner and classify class of target data based on this range. The main purpose of this paper is to know the effective classification method for software defect datasets that exploit information from the NASA MDP (PC1, CM1, JM1) datasets.