Evaluation of Feature Selection Techniques for Software Maintenance Prediction

Software quality is the ease with which a process or software fulfills customer's expectation. Software metrics have become an integral part of software development process to measure software quality quantitatively. However, not all the metrics affect quality in the same way. Selecting the relevant features reduces the time of computation and hence, improves efficiency. In this paper, different object-oriented metrics have been collected and then the most relevant metrics/features that affect maintainability in terms of CHANGE have been found using five feature selection techniques and three machine learning models. Additionally, the results have been compared with existing work and observed that results improve by selecting only the relevant features for prediction.