A Note on Static Software Reliability Models by GMDH: Comparison with a Multiple Regression Model

This paper provides a new approach to estimate the software reliability by the data sets obtained from the upper process of software development. The approach employs the group method of data handling (GMDH) which can be considered as an extension of a multiple regression analysis in this study. We first explain the concept of GMDH approach briefly and then, we discuss how the better model is constructed. By using the actual data sets from the upper process of the real software development, we try to estimate the number of inherent software faults. We compare two proposed models with a traditional multiple regression model in terms of the goodness-of-fit to the data sets.