A statistical comparison for evaluating the effectiveness of linear and nonlinear manifold detection techniques for software defect prediction
by Soumi Ghosh; Ajay Rana; Vineet Kansal
International Journal of Advanced Intelligence Paradigms (IJAIP), Vol. 12, No. 3/4, 2019

Abstract: Most of the software systems are released without predicting defects and therefore, this paper presents a new effective technique - manifold detection technique (MDT) is essential and different than earlier applied defect prediction methods like regression, feature selection methods, etc. In this paper, performance of classifiers has been compared with or without MDTs to evaluate the effectiveness of different MDTs (linear and nonlinear) by reducing the dimensions of software datasets. In this process, eight classifiers were applied to four PROMISE datasets to determine the best performing classifier with respect to prediction performance measuring factors (accuracy, precision, recall, F-measure, AUC, misclassification error) with or without MDTs. The experimental results statistically tested by paired two-tailed t-test proved that FastMVU is the most accurate result producing technique as compared to all other nonlinear MDTs and Bayesian network (BN) is the most effective technique for software defect prediction using with or without MDTs.

Online publication date: Thu, 28-Mar-2019

The full text of this article is only available to individual subscribers or to users at subscribing institutions.

 
Existing subscribers:
Go to Inderscience Online Journals to access the Full Text of this article.

Pay per view:
If you are not a subscriber and you just want to read the full contents of this article, buy online access here.

Complimentary Subscribers, Editors or Members of the Editorial Board of the International Journal of Advanced Intelligence Paradigms (IJAIP):
Login with your Inderscience username and password:

    Username:        Password:         

Forgotten your password?


Want to subscribe?
A subscription gives you complete access to all articles in the current issue, as well as to all articles in the previous three years (where applicable). See our Orders page to subscribe.

If you still need assistance, please email subs@inderscience.com