562015-03-24 , 法政大学大学院理工学・工学研究科
It is known that an RBF (Radial Basis Function) network model can be used to predict the number of inherent software faults from the data sets concerning software development data in the literature. To grasp the information on the number of software faults before beginning the testing process, can be the great help for the software development managers. The aim of this study is to provide the improved RBF network model in terms of the prediction accuracy.Therefore we propose a new RBF network model considering the correlation among the input data. In particular, we investigate the usefulness of introducing the correlation factors into the RBF networks and compare the performances between the proposed model and the previous one by using the actual data sets.