ENHANCING SOFTWARE QUALITY THROUGH MACHINE LEARNING- BASED DEFECT PREDICTION FOR ECONOMIC GROWTH AND DIGITAL INCLUSION

ABSTRACT

The increasing reliance on software systems in modern economies necessitates the development of reliable and high-quality software to support sustainable economic growth and digital inclusion. Meeting end-user requirements and expectations during the initial stage of software deployment is one of the most difficult tasks in the software development process. Therefore, detecting defects before deployment helps deliver high-quality software products and minimizes development costs. The research aims to identify and predict potential software defects at early stages of the software development lifecycle, thereby reducing maintenance costs, improving reliability, and accelerating software delivery. In this study, multiple machine learning classification algorithms such as Support Vector Machines (SVM), Naïve Bayes (NB), Random Forest (RF), Ada Boost (ADT) and Ensemble techniques are applied to the JMI dataset from NASA promise repository to evaluate their effectiveness in defect prediction. Data preprocessing techniques, correlation feature selection, and optimization are employed to improve prediction accuracy and reduce false positives. The performance of the models is assessed using standard evaluation or confusion metrics such as accuracy, precision, recall, and F1-score. The findings demonstrate that machine learning-based defect prediction significantly enhances software quality by enabling proactive identification of fault-prone modules. This contributes to efficient resource allocation, reduced development costs, and improved software reliability. This research provides valuable insights for software engineers, researchers, and policymakers on leveraging machine learning techniques to improve software quality and support inclusive digital transformation.

Keywords: Software Quality, Defect Prediction, Machine Learning, Machine Learning Classifiers, Digital Inclusion.