Required Characteristics for Software Reliability Growth Models

Software reliability growth models are used to estimate and predict software quality. Many software reliability growth models (SRGM) have been developed in the literature. It is a key issue how to appropriately select them since an inappropriate SRGM can give unreasonable prediction. This paper deals with this issue. It first discusses some characteristics for a SRGM to have, then presents five flexible SRGMs, and finally discusses how to determine the best SRGM from a list of candidate models for a given set of data. The usefulness of the suggested models and method are illustrated by a real-world example. The results show that all the suggested models outperform the well-known exponential model in terms of goodness-of-fit and predictive capability.