摘要
准确的软件成本预测对于投资决策和开发管控均具有重要意义。文中提出了一种基于灰关联与改进LS-SVM的软件成本预测方法,该方法利用灰色关联度与信息熵生成预测样本子集,降低了弱相关属性和冗余属性对预测的影响。基于LS-SVM构建成本预测模型,提高了小样本与非线性条件下的预测性能,并采用量子粒子群优化LS-SVM的模型参数,实现了软件成本的高性能估算。实验结果表明,该方法具有较高的预测精度,且泛化能力较强。
Accurate software cost prediction is of great significance for investment decision and development control.A software cost prediction method is proposed based on grey correlation and improved least squares support vector machine(LS-SVM).Grey correlation degree and information entropy is used to generate prediction sample subset,which reduces the influence of weak correlation attribute and redundant attribute on prediction.The cost prediction model is constructed based on LS-SVM,which improves the prediction performance under small sample and nonlinear conditions.The model parameters of LS-SVM are optimized by quantum particle swarm optimization to realize high performance estimation of software costs.Experimental results show that the method has high prediction accuracy and strong generalization ability.
作者
闫吉庆
黄贺
YAN Ji-qing;HUANG He(China Shenhua International Engineering Co.,Ltd.,Beijing 100007,China)
出处
《信息技术》
2020年第9期138-142,共5页
Information Technology
关键词
软件成本预测
灰色关联度
最小二乘支持向量机
量子粒子群
software cost prediction
grey correlation degree
least squares support vector machine
quantum particle swarm
