Comparing Efficiency of Software Fault Prediction Models Developed Through Binary and Multinomial Logistic Regression Techniques

Software fault prediction method used to improve the quality of software. Defective module leads to decrease the customer satisfaction and improve cost. Software fault prediction technique implies a good investment in better design in future systems to avoid building an error prone modules. The study used software metrics effectiveness in developing models in 2 aspects (binary and multinomial) Logistic Regression. We are developing multivariate (combined effect of object-oriented metrics) models in both aspects for finding the classes in different error categories for the three versions of Eclipse, the Java-based open-source Integrated Development Environment. The distribution of bugs among individual parts of a software system is not uniform, in that case Multinomial aspects helps the tester to prioritize the tests with the knowledge of error range or category and therefore, work more efficiently. Multinomial models are showing better result than Binary models.