基于灰色模型的工程造价指数组合预测模型构建  

Construction of a Combined Prediction Model for Engineering Cost Index Based on Grey Model

作　　者：李昌建 于海波 Li Changjian;Yu Haibo(Jiangsu Power Transmission and Transformation Co.Ltd,Nanjing Jiangsu 210028,China)

机构地区：[1]江苏省送变电有限公司,江苏南京210028

出　　处：《现代科学仪器》2024年第1期176-181,共6页Modern Scientific Instruments

摘　　要：随着大数据及人工智能的发展,构建合理的工程造价指数是工程造价发展的必然趋势。研究基于灰色预测模型和梯度提升决策树(Gradient Boosted Decision Tree,GBDT)预测模型,结合Stacking策略进行模型的组合,得到GM-GBDT工程造价指数组合预测模型。对模型的性能进行分析,发现三种模型中预测性能从高到低依次是GM-GBDT集成预测模型、GBDT预测模型、GM(1,N)预测模型,GM-GBDT集成预测模型对2020年1-12月工程造价指数的真实值和预测值的相对误差为3.86%-1.05%,平均相对误差为2.60%。实证分析结果表明,GM-GBDT联合模型,具有更好的整体预测能力,能够在GM(1,N)和GM的基础上,进一步提升预测准确率。With the development of big data and artificial intelligence,constructing a reasonable engineering cost index is an inevitable trend in the development of engineering costs.Research is based on the combination of grey prediction model and Gradient Boosted Decision Tree(GBDT)prediction model,combined with the Stacking strategy to obtain the GM-GBDT engineering cost index combination prediction model.Analyzing the performance of the models,it was found that among the three models,the highest to lowest predictive performance was the GM-GBDT integrated prediction model,GBDT prediction model,and GM(1,N)prediction model,The relative error between the actual and predicted values of the engineering cost index from January to December 2020 using the GM-GBDT integrated prediction model is 3.86%-1.05%,with an average relative error of 2.60%.The empirical analysis results indicate that the GM-GBDT joint model has better overall prediction ability and can further improve prediction accuracy on the basis of GM(1,N)and GM.

关 键 词：灰色模型 工程造价指数 组合预测 GM-GBDT 集成预测 

分 类 号：TU723[建筑科学—建筑技术科学]