Comparison of Classification Methods on Software Defect Data Sets

Nowadays it is difficult for us to imagine a lifewithout devices that is controlled by software. Softwarequality prediction is the important process of softwaredevelopment processes. It is a process of utilizingsoftware metrics such as code-level measurements anddefect data to estimate software quality modules. Amore useful and efficient mechanism is k NearestNeighbor method to classify class of target data basedon k nearest training dataset. By applying the concept ofk-NN, we propose a new mechanism called Class BaseWeighted k-NN with Biner Algorithm (CBW k-NN) tofind the range of training dataset where the target datahas the maximum likelihood of occurrence by Biner andclassify class of target data based on this range. Themain purpose of this paper is to know the effectiveclassification method for software defect datasets thatexploit information from the NASA MDP (PC1, CM1,JM1) datasets