Evaluation and comparison of software reliability growth models based on the nonhomogeneous Poisson process

The main aims of this paper are evaluate and to compare between several software reliability growth models (SRGMs) which are based on the nonhomogeneous Poisson process. The authors are motivated here by the recognition that although the number of existing SRGMs is large and getting steadily more numerous as researchers endeavour to inject more realism into their models, this increase is not matched by the amount of work which deals with the statistical analysis aspect or with the comparison between these models in their ability to produce reliable predictions. They use some well known plots such u-plot and y-plot, as well as statistical measures such as the prequential likelihood and Akaike Information Criterion to achieve our task.