Threshold design for cohesion and coupling metrics using K-means clustering Algorithm

shaitan singh meena, Satwinder Singh


As open source systems are becoming popular these days, there is more and more pressure on development teams to produce high-quality software. The quality of system (QoS) suffers even after undergoing testing; code smells still remain in the system. Various studies have been focusing on the relationship between metrics and code smells of class. Studies show that various metrics metric have been proposed to check the design quality of software during the development phase. A code smell is a suggestion or the description of an indication that something has gone wrong your code. Refactoring is done to clean the code and to decrease the chance of presenting the bugs in code. In this paper, describee is carried out to examine the relationship between cohesion and coupling metric and bad code smell for the open source code of ArgoUML four versions 0.24, 0.28, 0.30 and 0.34. Cohesion and coupling metrics design is done with the help of SourceMeter tool. PMD is used to collect all the bad code smell which are affecting the class. After that used technique k-means clustering, are used to find the threshold. In this work, it is proposed to use a clustering and metrics threshold based software bad code smell prediction approach and explore it on the dataset of ArgoUML versions. Clustering techniques have more importance in data mining especially when the data size is very large. Further, in this paper, the k-means algorithm has been applied for predicting code smells in program modules. Predicted models for one version are applied on other versions to check their accuracy. The results of the analysis are then used to conduct code smell classification based on the accuracy of one ArgoUML version and compared against the result of another version of ArgoUML.


Refactoring, Bad code smell, Threshold, Cohesion and Coupling metrics, k-means clustering, accuracy

Full Text:




  • There are currently no refbacks.

Copyright (c) 2017 International Journal of Advanced Research in Computer Science