FSDP: Frequent Software Defects Prediction Based on Defect Correlation Learning for Quality Software Development

Abstract

Software has become an essential and important part of every domain system. Developing quality software is critical to maintaining a stable and secure system. Most of the existing software defect prediction tasks focus on the various kinds of defects leftover in the software system, but they do not focus on the most common and frequent software defects that developers most commonly do and which have a significant effect on the quality of software development. These unnoticeable defects have a considerable impact on the functionality and also on development time, effort, and cost of the software. This paper propose a frequent software defects prediction (FSDP) mechanism based on defects Correlation learning method (CLM) utilizing various defects metrics. The aim of this work is to assist developers in accurately identifying software defects and support project managers in ensuring software quality by minimizing the presence of defect-prone code during development. The evaluation of FSDP was performed using NASA datasets in comparison with the conventional Naive Bayes, Support Vector Machine, and Random Forest classifiers and also compare with the state-of-the-arts methods to measure accuracy, precision, recall and F-Score to understand the impact of the prediction accuracy. The proposed FSDP achieves 98.88% accuracy and 98.86% precision in compare to state-of-the-arts classifiers methods indicates the effectiveness of the proposed approach in defect prediction.