Empirical Evaluation of Maintenance Effort Prediction using Inheritance and Interface-Oriented Class Complexity Metrics

Abstract

Objectives: The primary objective of this research is to estimate maintenance effort using complexity metrics and to demonstrate that the two selected metrics, CWICC and ICWCC, provide more accurate maintenance effort predictions compared to existing software complexity metrics. This will be validated through the application of various statistical techniques. Method: Two cognitive complexity metrics—Cognitive Weighted Inheritance Class Complexity (CWICC) and Interface-based Cognitive Weighted Class Complexity (ICWCC) were identified and empirically analyzed. Statistical techniques, including simple and multiple regression analysis, were applied to a collected dataset. The results were further compared with existing object-oriented metrics. Findings: The analysis demonstrates that cognitive complexity metrics can effectively predict software maintenance effort. The results also highlight differences in performance when compared to traditional object-oriented metrics, providing insights into their relative effectiveness. Novelty: This study introduces and validates two cognitive complexity metrics, CWICC and ICWCC to predict software maintenance effort. The novelty of this research work lies in establishing that the proposed metrics enable more accurate prediction of maintenance effort through closer alignment between predicted and actual maintenance time; for instance, ICWCC (160.54) and CWICC (159.09) closely approximate the observed value (162.08), outperforming WCC (139.32) and AWCC (154.29), thereby confirming their effectiveness as reliable predictors of software maintenance effort. Keywords: Maintenance, Effort, Cognitive Complexity Metrics, Maintenance Time, Correlation, Regression