Software Defect Classifier Using Linear Regression And Asymmetric Correlation Heatmap

This paper addresses the escalating complexity of software design and the rising frequency of defects, necessitating efficient defect analysis methods. Manual analysis becomes inadequate with increasing defect numbers, prompting the need for automatic defect classification. The study introduces defect management concepts and stresses the importance of defect classification in enhancing software quality and testing efficiency. Recent research has explored various methods for defect classification, yet the accuracy achieved often falls short of market demands. To address this gap, the paper proposes a comprehensive framework for defect prediction. The framework comprises data preprocessing, feature extraction via correlation-based analysis, and application of machine learning models like linear regression. Results demonstrate a promising accuracy of 98.7% with the proposed method. This paper explores the increasing complexity of software design and the growing need for efficient defect analysis and classification methods. It highlights the significance of defect analysis in improving software quality and reducing maintenance costs. Existing research on defect classification is summarized, emphasizing the gap between manual and automatic classification methods. In essence, the paper advocates for the use of software defect classification to enhance software quality, optimize testing resources, and reduce the effort involved in locating defects, ultimately leading to more efficient and reliable software development processes.