主成分分析法在居民出行生成模型中的构建研究

交通出行生成模型作为“四阶段法”出行的第一步，出行模型的精度将直接影响了整个四阶段法模型的结果；现行的居民出行生成预测模型主要有生成率法、交叉分类法、回归分析等。但是这些方法都没有从根本考虑到各种影响出行生成的变量之间的相互影响关系，而这种变量之间的相互关系往往会直接影响居民的出行特征，同时，常用的回归分析法中，由于最小二乘法的局限，往往会丢失变量或者出现负变量，而这些变量在交通出行中明显会产生出行量，这就和实际的交通出行特征有了背离。本文提出了在回归分析法之前首先采用主成分分析法来确定各影响变量之间的关系，确定主成分方程，进而在根据主成分方程和调查数据之间的回归分析来构建出行生成模型。通过这样的方法来有效的保护变量的非负性和体现变量之间相互关系对出行生成的影响。

Research on Principle Component Analysis in Trip Generation Model

As the first step in 4-stage model, the Trip Generation plays an important role in accuracy of transport model. Currently, Trip Rate, Cross-Category Analysis and Regression Analysis have been considered as the most popular methods in trip generation. Noticeably, all the above ways cannot sort out the relationship among the variables what will influence the result of trip generation. As a result, the paper has introduced the principle component analysis into trip generation. In detail, the coefficients among different variables in trip generation are going to be tested by Principle Component before the regression analysis is adopted in generation. In this way, the nonnegative parameter for variables and the relevance among variables will be guaranteed.