东南大学学报（自然科学版）2016，Vol.46Issue(6)：1313-1317,5.DOI:10.3969/j.issn.1001-0505.2016.06.034
基于潜变量S VM的出行方式预测模型
Forecasting model of travel mode based on latent variable SVM
摘要
Abstract
In order to improve the prediction accuracy of the travel mode choice model under small samples,a support vector machine (SVM)algorithm considering the low carbon travel psychologi-cal variables is proposed.Based on the theory of planned behavior (TPB),considering low carbon travel psychological factors,latent variable models with multiple causes and indicators are estab-lished.Substituting the forecasted latent variables into the SVM classifier,a SVM selection model with latent variables is then proposed.The mixed selection parameters are obtained using cross vali-dation optimization,and the model performance is validated based on urban residents' data in Yan-gtze River Delta region.Empirical results show that the established SVM selection model with latent variables has a better prediction accuracy,improved by 4.54%compared with the SVM without la-tent variables,and 2.56%by the traditional model with latent variables.Results prove that the mod-el still has a high precision with small samples.This study provides a theoretical reference for the travel choice model and low carbon travel choice research.关键词
混合选择模型/支持向量机/多原因多指标/计划行为理论/交叉验证算法Key words
mixed selection model/support vector machine(SVM)/multiple indicators and multi-ple causes/theory of planned behavior/cross validation algorithm分类
交通工程引用本文复制引用
陈月霞,陈龙,查奇芬,景鹏,谢君平,熊晓夏..基于潜变量S VM的出行方式预测模型[J].东南大学学报（自然科学版）,2016,46(6):1313-1317,5.基金项目
国家自然科学基金资助项目（71373105，61573171，51208232）、江苏省“六大人才高峰”资助项目（2015-JY-025）、江苏省高校科研创新计划资助项目（CXZZ12＿0663）． （）
