Welcome to the InfoSci Platform
Reference Hub2
Estimation of Target Defect Prediction Coverage in Heterogeneous Cross Software Projects

Estimation of Target Defect Prediction Coverage in Heterogeneous Cross Software Projects

Rohit Vashisht, Syed Afzal Murtaza Rizvi
Copyright: © 2021 |Volume: 12 |Issue: 1 |Pages: 21
ISSN: 1947-8186|EISSN: 1947-8194|EISBN13: 9781799861508|DOI: 10.4018/IJISMD.2021010104
Cite Article Cite Article

MLA

Vashisht, Rohit, and Syed Afzal Murtaza Rizvi. "Estimation of Target Defect Prediction Coverage in Heterogeneous Cross Software Projects." IJISMD vol.12, no.1 2021: pp.73-93. http://doi.org/10.4018/IJISMD.2021010104

APA

Vashisht, R. & Rizvi, S. A. (2021). Estimation of Target Defect Prediction Coverage in Heterogeneous Cross Software Projects. International Journal of Information System Modeling and Design (IJISMD), 12(1), 73-93. http://doi.org/10.4018/IJISMD.2021010104

Chicago

Vashisht, Rohit, and Syed Afzal Murtaza Rizvi. "Estimation of Target Defect Prediction Coverage in Heterogeneous Cross Software Projects," International Journal of Information System Modeling and Design (IJISMD) 12, no.1: 73-93. http://doi.org/10.4018/IJISMD.2021010104

Export Reference

Mendeley
Favorite Full-Issue Download

Abstract

Heterogeneous cross-project defect prediction (HCPDP) is an evolving area under quality assurance domain which aims to predict defects in a target project that has restricted historical defect data as well as completely non-uniform software metrics from other projects using a model built on another source project. The article discusses a particular source project group's problem of defect prediction coverage (DPC) and also proposes a novel two phase model for addressing this issue in HCPDP. The study has evaluated DPC on 13 benchmarked datasets in three open source software projects. One hundred percent of DPC is achieved with higher defect prediction accuracy for two project group pairs. The issue of partial DPC is found in third prediction pairs and a new strategy is proposed in the research study to overcome this issue. Furthermore, this paper compares HCPDP modeling with reference to with-in project defect prediction (WPDP), both empirically and theoretically, and it is found that the performance of WPDP is highly comparable to HCPDP and gradient boosting method performs best among all three classifiers.

Request Access

You do not own this content. Please login to recommend this title to your institution's librarian or purchase it from the IGI Global bookstore.