Identification of Code Smell Using Machine Learning

Abstract

Code smells are used to improve the quality of the software. Code smell detection detects the code problems like long method, large class, lazy class, long parameter list, feature envy, primitive obsession detector and too many literal detectors present in the code. In this work, two algorithms are used namely Support Vector Machine (SVM) and Random Forest algorithm. Support Vector Machine acts as classifier and the Random Forest algorithm are used for predicting the range of data. Decision making technique is used to identify the various problems present in the code. Code smell detection is a testing tool and it is mainly used by the developers, when the size of the code becomes unmanageable for manual detection. It also identifies deeper problems like syntax error, runtime error. It gives the output by analyzing the code in six different modules in this work.