abstract
- This chapter delves into how evolutionary algorithms (EAs) can be applied to identify code smells, framing it as a problem of selecting features in software code metrics. Code smells, which signal potential flaws in software design, typically demand labor intensive techniques for their detection. We investigate the efficiency of EA driven strategies that optimize the selection of features related to code quality, ultimately improving the precision and effectiveness of detecting code smells. We cover the basics of code smells and evolutionary algorithms, then detail our methodology for using EAs to select features that significantly improve code smell classification. Through experiments on industry-relevant software datasets, our approach is shown to improve traditional rule-based methods in detection accuracy. The chapter concludes with implications for future research and the potential of AI-enhanced solutions in software engineering. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2025.