摘要
针对出行生成预测中变量众多以及多重相关性问题,以常州调查数据为样本,分别采用逐步回归分析方法和偏最小二乘回归(PLSR)方法,建立基于土地利用形态的交通发生量模型。并对两者建模方法进行分析比较,结果表明PLSR能有效地解决变量多重相关性问题,模型意义清晰,效果显著。
There are serious multiple correlations of the urban land-use parameters which are the independent variables in the study of the relationships between land-use and traffic generation. The problem of multiple correlations of parameters is difficult to be solved by ordinary regression analysis methods. Taking the investigated parameters of traffic generation and land-use form of Changzhou city as a sample, this paper used partial least square regression （PLSR） and stepwise regression methods to establish the traffic generation model in combination with land-use form. Comparison between the two methods shows that the PLSR method is effective and reliable to solve the problem of multiple correlations of parameters and has clear physical meaning.
出处
《资源调查与环境》
2009年第4期285-290,共6页
Resources Survey & Environment
基金
中国地质调查局资助项目(121201010510218-5)
作者简介
张利（1984-）．女．江苏宿迁人．硕士研究生．主要研究方向为摄影测餐与遥感。
