On prediction of defect rates
Ana Maria Vladu, Ioana Făgărăşan · 2012
The prediction of software reliability can determine the current reliability of a product, using statistical techniques based on the failures data, obtained during testing or system usability. The major problem in predicting software reliability is their high complexity, as well as the excessive limitations of existing models. Therefore, choosing the appropriate reliability prediction model is essential. The purpose of this article is to study the defect prediction models, using the Rayleigh function. This function forecasts the defect discovery rate as a function of time through the software development process. Starting from an existing static prediction model, we have introduced a new approach, using the time-scale. Also, we have considered two methods for computing the model parameters, and have compared them to the real data set. The results have validated this model as a good estimation of the defect rate behavior. We will also suggest possible directions for extending the paper in the future.