Scholarlink

In recent years, software maintainability has become a critical attribute in software engineering to determine software quality. Hence, predicting this maintainability in an accurate and timely manner is a fundamental requirement for effective management during the software maintenance phase. This has led software developers to pay more attention to those modules that need high maintenance. The current study proposes a Learning Machine (ML) algorithm for Software Maintainability Prediction (SMP) using the Dataset of Students’ Software project requirements. This research demonstrates that advanced machine learning classification techniques, specifically Random Forest, AdaBoost and Voting Classifier, significantly enhance the accuracy of software maintainability assessments related to security requirements. For comparative analysis, Random Forest, AdaBoost and Voting Classifier demonstrate substantial accuracy gains over base models, achieving up to 87.85% in binary classification and 99.37% in multi-class classification, compared to the base models' maximum accuracies of 79.43% and 85.08%, respectively. These results highlight an effectiveness of advanced ML techniques in improving the accuracy of software maintainability assessments, underscoring their potential in better addressing security needs in software projects.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Enhancing Software Maintainability Through Machine Learning Classification Technique for Security Requirements

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Maitri Manya,
  • Rajkumar Sharma,
  • Vivek Richhariya

摘要

In recent years, software maintainability has become a critical attribute in software engineering to determine software quality. Hence, predicting this maintainability in an accurate and timely manner is a fundamental requirement for effective management during the software maintenance phase. This has led software developers to pay more attention to those modules that need high maintenance. The current study proposes a Learning Machine (ML) algorithm for Software Maintainability Prediction (SMP) using the Dataset of Students’ Software project requirements. This research demonstrates that advanced machine learning classification techniques, specifically Random Forest, AdaBoost and Voting Classifier, significantly enhance the accuracy of software maintainability assessments related to security requirements. For comparative analysis, Random Forest, AdaBoost and Voting Classifier demonstrate substantial accuracy gains over base models, achieving up to 87.85% in binary classification and 99.37% in multi-class classification, compared to the base models' maximum accuracies of 79.43% and 85.08%, respectively. These results highlight an effectiveness of advanced ML techniques in improving the accuracy of software maintainability assessments, underscoring their potential in better addressing security needs in software projects.