Artificial Neural Network Applications for Software Reliability Prediction

Chapter 5Identification of Software Fault-Prone Modules

It is important to prioritize testing based on fault-proneness of modules to achieve high software reliability in lesser time. Therefore, identification of fault-prone modules is important. Identification of fault-prone (FP) modules requires an approach similar to fault prediction as fault information is unavailable until testing is not carried out. To predict fault-prone (FP) modules, a mathematical model is generally used which relates software metrics with software fault-proneness. To achieve better model efficiency and accuracy, it is necessary that the important metrics are identified based on their relation with module fault-proneness and then some artificial intelligence method is used to learn and establish impact of these metrics on module fault-proneness. If the number of metrics available becomes too large, then high dimensionality problem occurs which may lead to extensive computation and degradation in model performance. All the metrics are not found to be equally important for the software fault-prone (FP) module prediction. Moreover, some redundant information is present in various metrics. Therefore, it is important to use most potential metrics with reduction of redundant information to achieve better performance from models.

In this chapter, two models a) artificial neural network (ANN) b) artificial neural network with particle swarm optimization (ANN-PSO) models are presented for fault-prone software ...