Empirical Analysis Of Software Restructuring Practices Across Development Lifecycle Stages: A Multi-Dimensional Study Of Approaches, Tools, And Techniques
Tamanna Tyagi, Sakshi Garg · INTERNATIONAL JOURNAL OF AGRICULTURAL SCIENCES · 2025
Long-lived software systems accrue structural complexity and technical debt, degrading maintainability and elevating defect rates. Software restructuring—behavior-preserving reorganization of internal code artifacts—mitigates these issues by reducing complexity and improving modularity. We present an empirical study of targeted refactorings applied to two mature open-source systems (Apache Ant and JEdit) over three two-week sprints. We measure impacts on cyclomatic complexity, coupling-between-objects (CBO), maintenance effort (mean issue-resolution time), and defect density. Additionally, we introduce a Hybrid Restructuring Metric (HRM) that combines classical metrics with AI-predicted maintenance effort to rank refactoring candidates. Results show complexity reductions of 36–39 %, maintenance-effort savings of ≈ 41 %, and defect-density drops of 36–40 %. HRM-guided refactorings yield an extra 8 % improvement in maintenance effort over expert selection. All results are statistically significant (paired t-tests, p 1.2).