Review: Software fault prediction: A literature review and current trends
Expert Systems with Applications: An International Journal
A survey in the area of machine learning and its application for software quality prediction
ACM SIGSOFT Software Engineering Notes
| Hi-index | 0.00 |
Software quality prediction models are used toachieve high software reliability. Prediction models thatestimate a quality factor for software modules can beused in directing corrective efforts. Precise quantitative prediction values for the quality factor is often notsuffcient. Instead, predicting the rank-order of modules with respect to the quality factor may be morebeneficial to the development team. A module-ordermodel (MOM) uses an underlying quantitative prediction model to predict this rank-order.This paper compares performances of module-ordermodels of two different count models which are used asthe underlying prediction models. They are the Poisson regression model (PRM) and the zero-inflated Poisson (ZIP) regression model. It is demonstrated thatimproving a count model for prediction does not ensure a better MOM performance. A case study of afull-scale industrial software system is used to compareperformances of module-order models of the two countmodels. It was observe dthat improving prediction ofthe Poisson count model by using zero-inflated Poisson regression did not yield module-order models withbetter performance. Thus, it was concluded that the degree of prediction accuracy of the underlying model didnot influence the results of the subsequent module-ordermodel. Module-order modeling is proven to be a robustand effective method even though both underlying prediction may sometimes lack acceptable prediction accuracy.