Improving the Accuracy of the Littlewood-Verrall Model

A Bayes model of a system with stochastically monotonous failure rate, proposed by Littlewood & Verrall, has been tested in a simulated experiment. We show that the model can be improved by introducing pseudo-failures in order to enrich the input data. This leads to a greater accuracy in most predictions and makes the model more adaptive to the data.