Software life cycle-based defects prediction and diagnosis technique research

A model based on Bayesian network is put forward to predict and diagnose software defects before project. In the model causes and effects inference is used to predict defects, and the Bayesian formula is introduced to analyze the prediction result that helps to find the root reason of bring defects. The model considers every phase of software life cycle, such as requirement, design, development and testing, maintenance. The model computes predict result through variational weight of affect-factor. The computation results of specifically affect-factor by model compared with practical defect in same condition which to indict that the model is validity. The model can predict and find defects early and effectively, it can control the software quality and the cost of development.