PREDICTING SOFTWARE BUGS OF NEWLY AND LARGE DATASETS THROUGH A UNIFIED NEURO-FUZZY APPROACH: RELIABILITY PERSPECTIVE

Abstract

Reliability of software is an essential concern for users for a long time. Software reliability is mainly obtained through modeling and estimating. There are numerous methods for reducing the failure rate. However, the existing methods are nonlinear. Hence the parameter estimation of these methods is difficult. This paper concerns on estimation and prediction of software reliability through different soft computing methods for improving the reliability of software. For estimation and prediction, the authors of this paper take two soft computing methodologies, including fuzzy logic and neural network. The outcomes seem to give satisfactory results on large datasets. For experiments, this paper is using two different large datasets of Apache server and MyLyn application software for showing the effectiveness of the results. The proposed methods of prediction would be useful for practitioners to simplify the procedures during software development in large datasets for reducing failures of software.