THE IMPROVEMENT OF FITTING QUALITIES BY USING NHPP MODELS
Z Iong · 1992
It is easy to fit the cumulative error data collected during the software test phase by using NHPP models.But the fitted results of a single NHPP model are often unsatisfactory because the model assump- tions do not fit the total cumulative error curve.In addition,because of the lack of information in an incomplete data set,the results produced by using a single NHPP model will be affected.In this paper, we discuss the improvement methods including the application of EM algorithm,and piecewise fitting process.In order to compare the different fitting results,we define the fitability rate numerically in this paper.