Enhancing software reliability prediction based on hybrid fuzzy K-nearest neighbor with glowworm swarm optimization (FKNN-GSO) algorithm

Abstract

Predicting software reliability means gauging the future occurrences of failures in software in order to align the process of the software maintenance. This paper presents a model based on FKNN (Fuzzy k-Nearest Neighbor) and nature inspired Glowworm swarm optimization (GSO) to understand the relationship between the data of software failure time and the nearest n failure time and finally predict the reliability of the software. Glowworm-Swarm Optimization (GSO) is used to search finest combination of weights aimed to obtain maximum regression accuracy and fuzzy k-nearest neighbor (FKNN) to allocate the degree of membership to various software metrics using fuzzy logic concepts. The performance of the proposed model has been compared with the known existing models to evaluate the prediction efficiency of GSO- FKNN.