广义回归神经网络在软件质量预测中的应用  

An Application of General Regression Neural Network to Software Quality Prediction

作　　者：马慧敏[1] 宋雨[1] 许正伟[1] 

机构地区：[1]华北电力大学计算机科学与技术学院,河北保定071003

出　　处：《计算机工程与应用》2004年第29期217-219,共3页Computer Engineering and Applications

摘　　要：软件质量预测技术是软件质量评价体系中的关键技术,它能够对用户所关心的软件质量特性进行评价。广义回归神经网络在逼近能力、分类能力和学习速度方面具有较强优势。采用基于软件度量的广义回归神经网络构造质量预测模型能够从历史数据中寻找软件度量之间的相关关系。对软件缺陷数进行预测的实验说明了模型的有效性、精确性,实验结果令人满意。Software quality prediction technology is the key technology of software quality evaluation system and can evaluate software quality characteristic concerned by customers.General regression neural network is proved with certain superiority in the ability of approaching,classification and learning speed.Adopting GRNN based on software metrics to construct quality prediction model can find correlativity between software metrics from history data.The experimental results of prediction of number of faults are satisfactory and indicate that the model is effective and accurate.

关 键 词：软件质量预测 广义回归神经网络 软件度量 缺陷数 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]