基于偏最小二乘回归法的城市土地利用与交通发生量关系模型研究  

Research on Model of Relationship between Urban Land-use and Traffic Volume Based on Partial Least Squares Regression

作　　者：戚浩平[1] 张利[1] 王炜[1] 陆建[1] 

机构地区：[1]东南大学交通学院,江苏南京210096

出　　处：《公路交通科技》2011年第3期138-142,147,共6页Journal of Highway and Transportation Research and Development

基　　金：国家重点基础研究发展计划(九七三计划)资助项目(2006CB70551)

摘　　要：以实际调查得到的建筑容积率、建筑密度、总建筑面积、交通小区面积以及各类型土地利用面积等14个土地利用指标作为直接变量,建立交通发生量预测模型。相关分析的结果表明土地利用指标间存在着严重的多重相关性,普通的回归分析方法很难有效解决这一问题。采用偏最小二乘回归方法,通过构建主成分,采用交叉有效性检验确定成分个数,最终建立的交通发生量预测模型物理意义明确,精度较好。研究结果表明:居住用地、建筑容积率、总建筑面积等土地利用指标对交通发生量的影响最大。Fourteen urban land-use indexes from investigation,which can indicate the land-use intensity such as building volume rate,building density,overall floorage,area of traffic zone and each land use type,were used as direct variables to establish the forecasting model of traffic volume.Correlation analysis shows that there are serious multiple correlations in the urban land-use indexes.The question is difficult to be solved by ordinary regression analysis methods,but partial least squares regression（PLSR）method can solve it effectively.The uncorrelated principal components which can pass through the cross-validation test were used to establish the forecasting model of traffic volume.The results show that residential estate,building volume rate and overall floorage have marked influence to traffic volume.

关 键 词：交通工程 预测模型 偏最小二乘回归法 土地利用 交通量 

分 类 号：U491[交通运输工程—交通运输规划与管理]