Heterogeneous Cross Project Defect Prediction – A Survey

Abstract

In the testing phase of Software Development Life Cycle (SDLC), Software Defect Prediction (SDP) is one of the pivotal task which finds the modules that are more vulnerable to defects and therefore need substantial testing for the early identification of these defects. A lot of work has been done on Cross - Project Defect Prediction (CPDP) that aims to predict defects in the target project lacking in historical defect prediction data or having limited defect data to build an effective generalized defect prediction model. Mostly, CPDP approaches predict the defects in target project on the basis of similar metrics found between source and target project. This paper focuses on the prediction of defects using a Heterogeneous metric set such that no common metrics exist between the source and the target projects. In this paper, a systematic literature study has been done to quote the main findings about CPDP from year 2002 to 2019. The main purpose of this survey is to put forward the adequate content in front of computer science researchers for exploring the specific area and to provide various future directions in this field.