Recommending Extract Method Opportunities Using Machine Learnig

Refactoring is a technique for restructuring an existing body of code, altering its internal structure without changing its external behavior. It is a very important activity to improve software maintainability and readability. In this research, we propose a machine learning approach for suggesting targets of extract method refactoring using source code features. As an experiment to evaluate proposed approach, we applied proposed approach to five open source software. From the result of the experiment, from 60% to 90% of refactored methods can be correctly predicted using the prediction models built by proposed approach. In addition, the number of statements and the degree of cohesion in a single method are strongly related to whether the method is refactored or not.