Maintainability prediction of web service using support vector machine with various kernel methods

Abstract

The present day software are mostly developed based on Service-Oriented Computing (SOC), which assembles loosely coupled pieces of software called services. With the increase in the number of development of these varieties of service oriented software, their effective maintenance plays an important role for the developers. The quality of SOC can be best assessed by the use of software metrics. In this paper, different object-oriented software metrics have been considered in order to design a model for predicting maintainability of SOC paradigm. Further support vector machine with different type of kernels have been considered for predicting maintainability of SOC paradigm. This paper also focuses on the effectiveness of feature selection techniques such as univariate logistic regression analysis, cross correlation analysis, rough set analysis, and principal component analysis. The results show that, maintainability of SOC paradigm can be predicted by application of various object-oriented metrics. The results further indicated that, it is possible to find a small subset of object-oriented metrics out of total available various object-oriented metrics, that enables prediction of maintainability with higher accuracy.