Abstract
The evaluation of reliability is very important in developing software efficiently. This paper describes the construction of a software developmentβassist tool, which executes the reliability evaluation of the software quantitatively in the testing phase. In this tool, the following models are employed for the analysis of the software error data. The stochastic models based on nonhomogeneous Poisson processes are exponential, delayed Sβshaped and inflection Sβshaped software reliability growth types. The deterministic regression models are logistic and Gomperz curve models. The procedure for evaluating the software reliability is as follows: The data observed in the testing phase are read in. The unknown parameters in each model are estimated, and the goodnessβofβfit test of the estimated result is examined. The results of the estimation for the reliability evaluation measures as well as the observed data are displayed in the form of curves. The forementioned procedures are implemented in a program package, using BASIC language operating on MSβDOS. Using the software reliability evaluation tool developed in this study, an executing example is presented based on the actually observed software error data.