基于支持向量机的面向对象软件易发性故障预测

Object-Oriented Software Fault-Proneness Prediction Using Support Vector Machine

本文考虑软件故障严重程度,并采用C&K面向对象度量集,以支持向量机分析方法为数学工具,建立一种基于面向对象软件易发性故障预测模型。实验结果表明,与基于朴素贝叶斯的预测模型、随机预测模型和NNge预测模型相比,本文提出的预测模型对于高严重程度故障、低严重程度故障以及未划分故障严重程度的情形均获得较好的预测效果。

With software fault severity considered, a software fault-proneness prediction model is proposed in this paper by Support Vector Machine and the Chidarnber-Kemerer （C＆K） object- oriented metrics. The experimental results show that this presented model obtains better results than that of the Naive Bayesian, the Random forest, and the NNge predietion models when the high and low severity faults and the ungraded severity fault are distinguished.