Burr-type NHPP-based software reliability models and their applications with two type of fault count data

In this paper, we summarize the so-called Burr-type software reliability models (SRMs) based on the non-homogeneous Poisson process (NHPP) and comprehensively evaluate the model performances by comparing them with the existing NHPP-based SRMs. Two kinds of software fault count data are considered; fault-detection time-domain data and fault-detection time-interval data (group data). For 8 data sets in each fault count type, we estimate the model parameters by means of the maximum likelihood estimation and evaluate the performance metrics in terms of goodness-of-fit and prediction. It is shown that the Burr-type NHPP-based SRMs could show the better performances than the existing NHPP-based SRMs in many cases. The main contribution of the paper consists in suggesting that the Burr-type NHPP-based SRMs should be the possible candidates for selecting the best SRM in terms of goodness-of-fit and .