Software Defect Prediction Accuracy using ML

Software defect detection remains a fundamental challenge in modern software development, with traditional approaches often suffering from high false positive rates and limited scalability across diverse codebases. This research introduces HSDDF (Hybrid Software Defect Detection Framework), an innovative system that combines machine learning algorithms with enhanced static code analysis to achieve superior defect identification accuracy. Our methodology integrates Support Vector Machines, Decision Trees, and Naive Bayes classifiers with advanced static analysis metrics including cyclomatic complexity, code coverage patterns, and dependency analysis.