(A) systematic class refactoring approach based on metrics

Abstract

Refactoring has been considered an effective technique for improving the maintainability of software. However, refactoring is a difficult task because it is hard to determine where to apply which refactoring operations. Since refactoring usually relies on human intuitions, it is harder especially troublesome for the beginners. In this thesis, we propose a systematic refactoring approach based on metrics in order to help the refactoring process. We employ a number of metrics both for identifying refactoring candidate classes and for selecting suitable refactoring operations. We also suggest a refactoring process that will help apply refactoring systematically. We apply our refactoring approach to a program of moderate size and show that our approach is effective for locating refactoring candidate classes and choosing appropriate refactoring operations.