A Review on Software Fault Detection using a Classification Model with Dimensionality Reduction Technique

Abstract

Software plays the most important role in every organization it requires high-quality software. If a fault happens in this system then it causes high financial costs and affects people's lives. So, it is important to develop fault-free software. Sometimes, a single fault can cause the entire system to frailer. So in the SDLC life cycle Fault prediction at an early stage is the most important activity it helps in effectively utilize the resources for better quality assurance. So before delivering the software to market it is important to identify defects in the software because it increases the customer satisfaction level. here in this survey paper present an ensemble approach to identifying fault before delivering the software. Ensemble classifier improved classification performance compared to the single classifier. So improved the accuracy the new algorithm is proposed that is "improved random forest" it works with random forest classifier with filter-based feature selection method. The feature selection method reduces the dimensionality and selects the best subset of features and gives that subset to the random forest classifier. The experiment carried the public NASA dataset of the PROMISE repository.