Int J Performability Eng ›› 2023, Vol. 19 ›› Issue (1): 33-42.doi: 10.23940/ijpe.23.01.p4.3342

Previous Articles     Next Articles

Maintainability of Service-Oriented Architecture using Hybrid K-means Clustering Approach

Arvind Kumar Mishraa,*, Renuka Nagpalb, Kirti Sethc, and Rajni Sehgalb   

  1. aAmity Institute of Information Technology, Amity University, Noida, 201303, India;
    bAmity School of Engineering and Technology, Amity University, Noida, 201303, India;
    cInha University in Tashkent, Tashkent, 100170, Uzbekistan
  • Contact: *E-mail address:

Abstract: The process of making changes to software after it has been delivered to the client is known as maintainability. Maintainability deals with new or changed client requirements. Service-oriented architecture (SOA) is a method for developing applications that helps services work on different environments. SOA works on patterns of distributed systems that help different applications communicate with each other using different protocols. To assess the maintainability of service-oriented architecture, different factors are required. Some of these factors are analyzability, changeability, stability, and testability. Modification is the process of upgrading the software functionality. After modification of service-oriented architecture, the module will go to the testing phase. The evaluation and verification of whether a software product or application performs as intended is known as testing. The testing phase is a combination of various stages, such as individual module testing and testing after collaborations between them. This testing stage is time-consuming in the maintenance process. The term “outlier" refers to a module in software systems that deviates significantly from the rest of the module. It represents the collection of data, variables, and methods. For instance, the program might have been coded mistakenly or an investigation might not have been run accurately. To detect the outlier module, test cases are needed. A methodology is proposed to reduce the predefined test cases. K-means clustering is the best approach to calculate the number of test cases, but the outlier is not automatically determined. In this paper, a hybrid clustering approach is applied to detect the outlier. This clustering method is used in software testing to count the number of comments in various software and in medical science to diagnose the disease of Covid patients. The experimental outcomes show that our strategy achieves better results.

Key words: clustering approach, k-means clustering, outlier detection, data analysis, Euclidean distance