Abstract
Software fault prediction helps in early identification of software faults and as a result it improves the software quality. It uses previous software metrics and fault data as independent features, to detect whether there is a fault in the software or not. Early prediction of software faults saves a lot of money and effort required to correct those faults. But, as the amount of data is very huge, it is essential for feature selection to get the most useful information. In this paper, we proposed a Genetic Algorithm-based feature selection method that identifies the most useful subset of features for classification purposes. We used a combination of Genetic Algorithm with KNN Classifier, Decision Tree Classifier and Naive Bayes Classifier for our experiments. Our results suggest that, by using Genetic Algorithm for feature selection, our prediction accuracy improved in all the three classifiers for all the datasets and also the number of features were reduced.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have