Towards Formulating Dynamic Model for Predicting Defects in System Testing using Metrics in Prior Phases

Abstract

Many studies have been carried out in formulating software defect prediction but it is of limited knowledge that those studies emphasized on predicting defects in system testing phase. This study specifically focuses on establishing a prediction model for system testing defects by exploiting metrics prior to system testing under V-model development. The initiative helps independent testing team to prevent as many defects as possible from escaping to production environment. The proposed model analyzes development-related and testing-related metrics collected from requirement, design and construction phases in determining which of those could significantly predict defects at the start of system testing. By applying statistical analysis to those metrics, this model able to formulate one generalized mathematical equation for predicting defects in system testing. The model applies 95% prediction interval to ensure the accuracy of the prediction.