Research On A Method Of Software Reliability Prediction Model

Software defects seriously affect software reliability. Software reliability is accurately predicted, which can guarantee normal use of software. For the prediction of software reliability, a software reliability prediction model based on support vector regression is constructed. Firstly, software metrics are extracted. Then data is preprocessed by the method of normalization and random sequence, and it is divided into training sets and testing sets for regression prediction; Secondly, the grid search method is introduced to optimize parameters in the support vector regression model; Finally, through experimental comparison and analysis, it is proved that this prediction method can indeed greatly improve the prediction accuracy of software reliability.