Hybrid Bayesian Network Models for Predicting Software Reliability
Mark Blackburn, Benjamin Huddell · 2012
This paper discusses the results of applying a hybrid Bayesian Network to predict software reliability measures. The model combined quantitative testing data with subjective expert judgment about program-specific aspects over many releases. Six different programs were analyzed using historical data to validate the model. The model predictions varied from project-to-project suggesting that additional program variables should be included in the model.