Early software defects prediction using fuzzy logic

Abstract

This paper presents a model to predict the number of residual defects before testing phase. The developers need this information a priori for optimal testing resource planning and quality assessment of the software being developed. In the early stages of software development life cycle (SDLC), software residual defects are affected by both product and process characteristics of the project. These product and process characteristics are embedded in software metrics which have subjective assessments in the early stages of SDLC. Therefore, software metrics are considered for developing the model for early software defects prediction. This paper uses the software size metric and three metrics of requirement analysis phase for predicting the residual defects that are likely to be found during testing or operational usage using fuzzy logic. The predictive capability of the proposed approach is examined using qualitative data of requirement metrics of twenty real software projects and results are compared with the existing model. © RAMS Consultants.