Enhancing Quality Assurance with Machine Learning: A Predictive Approach to Defect Tracking and Risk Mitigation

Abstract

Ensuring high software quality is vital in today’s fast-paced development environment. Traditional Quality Assurance (QA) approaches have shown limitations, especially when addressing defects reactively. This paper extends the work of Kothamali and Banik (2019) on machine learning (ML) algorithms in QA, comparing predictive defect tracking methodologies and examining the impact of different QA metrics on risk management. The study employs supervised and unsupervised learning models to identify defect-prone areas and manage software vulnerabilities more effectively. A comparative analysis of logistic regression, random forests, support vector machines (SVM), and neural networks is performed to determine which algorithm best predicts defects in various scenarios. This research aims to provide a holistic understanding of how machine learning can revolutionize QA practices, leading to more efficient defect management and risk assessment.