Design and Analysis of Improvised Genetic Algorithm with Particle Swarm Optimization for Code Smell Detection

Abstract

Software development phase is very important in the Software Development Life Cycle. Software maintenance is a difficult process if code smells exist in the code. The poor design of code development is called code smells. The code smells are identified by various tools using various approaches. Many code smell approaches are rule based. The rule based approaches are based on trial and error method. Genetic Algorithm is a heuristic Algorithm by Darwin’s Theory. This paper presents a metric based code smell detection approach by Genetic Algorithm with particle swarm optimization based on Euclidean data distance. The Euclidean data distance gives best proximity value between two points. Our approach is evaluated on the three open source projects like JFreeChart v1.0.9, Log4J v1.2.1 and Xerces-J for identifying the eight types of code smells namely Functional Decomposition, Feature Envy, Blob, Long Parameter List, Spaghetti Code, Data Class, Lazy Class, Shotgun Surgery.