Software Fault Detection using Machine Learning Algorithm |
Author(s): |
Pooja Garg , DIET, KARNAL; Mr. Bhushan Dua |
Keywords: |
Support Vector Machines (SVM) Bootstrapping, Clustering, Random Projection, Software Metrics, Metrics datasets |
Abstract |
Software fault prediction indicates the likelihood of software fault at an early stage of software development process and hence it will be easier to identify and correct them and also reduce faults that would occur at later stages. This will improve the overall quality of the software product. In the recent years, several machine learning techniques which uses examples of faulty and non-faulty modules to build prediction models. Software metric have been used as input to these machine learning techniques to represent the software modules. Support Vector Machines (SVM) is the main algorithm which has been used for classification of faulty and non-faulty modules. But prior to using SVM, A few data pre-processing methods has been proposed such as bootstrapping, clustering and random projection. These methods are needed because of certain problems with metric datasets such as class imbalance, noise, small dataset size and high dimension. It has showed by experiments that when software metrics dataset is pre-processed and transformed using above techniques, the performance of SVM in predicting faulty and non-faulty modules is better. The accuracy measure used for comparing performances of different models was F-measure. F-measure has been used because of its robustness to class imbalance. For some models F-measure has increased by about 40%, which is very encouraging. Experimental results shows that the proposed approach worked better than existing approach using MATLAB and Weka programming. |
Other Details |
Paper ID: IJSRDV3I40991 Published in: Volume : 3, Issue : 4 Publication Date: 01/07/2015 Page(s): 3475-3480 |
Article Preview |
|
|