Predicting the Maintainability of Object Oriented Software Using Design Metrics – An Evolutionary Case Study of Open Source Software

Software maintainability is an important external quality attribute that contributes significantly when the quality of any software is evaluated. There has been several proprietary software systems which have evolved using the Object-Oriented paradigm and several research works have investigated ways in bettering the maintainability levels of such software. Recently, many open source software systems have also started evolving the Object-Oriented way and there is a need to investigate these software systems on how they behave with respect to maintainability. Many empirical studies have been conducted in the recent past to indicate that software design metrics help in better prediction of maintainability when compared to measuring it during later stages of the software development life cycle. The objective of this research is to perform an empirical study on the evolution of a popular open source software JFreeChart by analyzing a few package design metrics like size, coupling and stability metrics proposed by the Martin suite. This work analyzes the impact of these metrics on the evolution of JFreeChart and proposes a predictive model to evaluate its maintainability. The proposed predictive model has been found to provide a very high prediction accuracy of 90.1%