Enhancing Software Maintainability Prediction Using an Optimizable-Support Vector Machine

Maintainability is a crucial phase in any software system, it is not just fixing bugs but strategic planning for long-term success of any software system. Software maintainability is influenced by many factors, and there are several ways for predicting it like metrics extraction, test coverage, software complexity measures, machine learning (ML) models, anti-detection pattern and natural language processing, etc. As maintainability is a complex term, effectively predicting it requires a thorough understanding of a wide range of software attributes and development techniques. We proposed feature selection-ranking based optimizable support vector machines (SVMs) and applied with feature selection ranking algorithms. Our proposal is performing better with validation accuracy of 87.9%.