Machine Learning Powered Code Smell Detection as a Business Improvement Tool

Code smell represents the level of human interpretability in a software project, which becomes increasingly challenging as modern-day software projects grow in complexity. Machine learning has promising signs of solving the problem of code smell detection but will ultimately be limited by the training dataset of the model. This paper investigates some machine learning approaches for code smell detection, the implications of using such a system, integration with business processes and how such a system would fit into IT governance using Latvia as an example.