A Novel Ensemble Learning for Defect Detection Method With Uncertain Data

A Novel Ensemble Learning for Defect Detection Method With Uncertain Data

Sreedevi E., PremaLatha V., Prasanth Y., Sivakumar S.
Copyright: © 2021 |Pages: 13
DOI: 10.4018/978-1-7998-3335-2.ch005
OnDemand:
(Individual Chapters)
Available
$37.50
Current Special Offers
No Current Special Offers
TOTAL SAVINGS: $37.50

Abstract

Data which contains noise is termed as uncertain data, and the presence of noise makes a deviation in the correct, intended, or original values. Size and complexity of the software products are the two main reasons for uncertain data set that identifying defective modules in uncertain datasets has become a challenging issue. In this chapter, the authors implemented a multi-learner ensemble model for uncertain datasets for defect detection. In this model, different weak classifiers are optimized to improve the classification rate on uncertain data. They have implemented their proposed model on NASA(PROMISE) metric data program repository. Accuracy is used as performance evaluation metric for our multi-learner ensemble defect detection model and ensemble model outcome achieved higher accuracy rate of 97% and when compared to another classification model.
Chapter Preview
Top

To search the stochastic process regarding the defect factors and discover the interval between the variable level the defect prediction models are calculated (Zimmermann & Nagappan, 2009), (Ceylan et al., 2006). To calculate the total amount of defects that appear during the defect dependency test they are calculated by utilizing non-homogeneous poisson process. Search the poisson process P (t) for each one defect, the probability of obtaining k defects at the time and it is declared in time of the Poisson distribution with mean m(t) as

Prob(P(t) = k) = m(t)n.e-ml/n!

Complete Chapter List

Search this Book:
Reset