Software metrics play a key role in measuring attributes that are important for the success of a software project. Measurements of these metrics tell us various key aspects of system. This in turn supports knowledgeable decision making by which we can enhance the quality of system. Maintenance is a process of revisions or corrections made to software systems after their first release. The key feature of software development is change. Hence it is important to develop software that is easy to modify and is thus maintainable. This paper evaluates the existing Oman and Hagemeister maintainability index model which calculates maintainability index (MI) based on Cyclomatic Complexity, Lines of code and Halsted volume. For this purpose, software metric datasets of Lucene, which is open source software of 163085 lines of code are used, and it is shown that the existing Oman and Hagemeister maintainability index mode model is not a good a predictor of maintainability. A new maintainability index model is proposed with a new set of predictor metrics. The new proposed model is a marked improvement over the existing Oman andHagemeister maintainability index model. The coefficient of determination ( ) of the new proposed maintainability model is 0.984 and correlation coefficient(R) is 0.992 as compared to the Oman and Hagemeister model whose correlation coefficient(R) is 0.320

No abstract