AI-Driven Predictive Analytics for Software Quality Improvement

Abstract

This paper presents an advanced machine learning-based predictive analytics framework aimed at enhancing quality assurance (QA) practices in large-scale software systems. Building on the foundational work of Kothamali and Banik (2019), which introduced a model utilizing machine learning algorithms for defect tracking and risk management, this study proposes a more refined and adaptive analytics model. The new framework integrates key factors, including defect history, change complexity, and test execution outcomes, to predict potential failure zones in software releases. By incorporating these elements, the system proactively identifies risk areas and enables more targeted intervention during the testing phase. To validate its effectiveness, the proposed system was tested across two distinct enterprise application environments, both showing a measurable increase in test efficiency and a significant reduction in issue resolution time. This practical validation underscores the model’s real-world applicability and its capacity to optimize the QA process. Furthermore, this work reinforces the foundational relevance of the model developed by Kothamali and Banik, demonstrating its continued utility in addressing modern QA challenges and advancing the state of software quality assurance in complex development environments.