MLR和SVM模型在水电工程价格指数预测中的应用

水电工程价格指数综合反映不同地区各年度投资变化趋势,价格指数的准确预测可为政府和业主宏观控制工程总投资提供参考数据,对造价控制尤为重要。以2005~2015年水电工程综合价格指数及各类调价因子为基础,建立多元线性回归(MLR)和支持向量机(SVM)价格指数预测模型,对比分析两模型的实效性。实证表明,当样本容量较小时,MLR预测模型更为适用,精度达99.73%,标准差低达0.002 9;当样本容量较大时,SVM模型对水电工程价格指数预测精度更高,达99.77%,可见所提方法对价格指数预测精度高、稳定性强,能为造价管理提供参考。

Application of MLR and SVM Model in Prediction of Hydropower Project Price Index

Hydropower project price index comprehensively reflect the tendency of annual investment in different regions. The accurate prediction of the price index can provide reference data for the government and the owner of the macro control project total investment, especially for cost control. In this paper, price index is predicted by MLR and SVM model based on the comprehensive price index of hydropower project and all kinds of price adjustment factors in the years from 2005 to 2015. And then the effectiveness of the two models is analyzed through comparison. Evidence shows that when the sample data is small, the MLR model is more suitable whose accuracy is 99.73% and the standard deviation is as low as 0.0029; when sample data is large, the SVM model is more accurate for prediction of hydropower project price index whose accuracy is 99.77%. Therefore, the proposed method has high accuracy and stability, which can provide reference for the cost management.