【摘要】 随着社会各行各业对软件开发投资的日益增长,产业界和学术界越来越关注可靠的软件成本估算,以有效控制软件开发过程中相关风险。为了能更准确地估算软件成本,提出一种带遗传算法优化参数的支持向量回归机模型,用遗传算法来优化支持向量回归机模型中的参数集(C,γ,ε),可以避免参数选择的盲目性,能显著提高支持向量回归机模型的预测能力。分别用IBM DP、Kemerer和Hallmark三个数据库来验证模型的有效性,并与常用的线性回归模型进行对比,结果显示采用遗传算法优化的支持向量回归机模型具有很好的学习精度和推广能力,在MMRE和Pred(0.25)两个标准上都优于线性回归模型。更多还原

【Abstract】 With software development becoming an essential investment for many organizations recently,both the software industry and academic communities have become more and more concerned about reliable and accurate estimates of software development effort.This paper examines the potentials of the software effort estimation model by integrating a genetic algorithm(GA)to the support vector regression(SVR)model with RBF kernels.The genetic algorithm(GA)method is adapted to optimize the parameters(Cγ,ε)in support vector regression.The experiments are carried out using three datasets of software projects from IBM DP,Kemerer and Hallmark.The effectiveness of the GA-SVR model is compared with other techniques such as linear regression in terms of the error measure which is mean magnitude relative error(MMRE)and Pred(0.25).The results have shown that GA-SVR model has superior learning accuracy and generalization performance and significantly outperforms linear regression in three datasets.更多还原