A Travel Mode Forecasting Model Based on Low-carbon Psychological Latent Variable Logit Model

Yue-xia CHEN,Long CHEN,Qi-fen ZHA,Peng JING,Xiao-xia XIONG

DOI: https://doi.org/10.3969/j.issn.1002-0268.2017.09.014

2017-01-01

Journal of Highway and Transportation Research and Development

Abstract:For in-depth analysing low-carbon travel behavior mechanism, a method with introducing low-carbon psychological variables into discrete choice model is proposed. Considering the low-carbon travel psychological factors based on the theory of planned behavior, the model of multiple indicators and multiple causes including latent variables is established. A hybrid selection model with low-carbon travel psychological variables which introduces forecasted latent variables into the multinomial Logit model is constructed. Taking the investigation samples of urban resident commuters in Zhenjiang City as the research object, the empirical result shows that (1) the built hybrid selection model has a better prediction accuracy, which is improved by 2. 45％ compared with the traditional discrete choice model;(2) the influences of low-carbon psychological variables on the choice of travel mode are different, the"behavior attitude" of low-carbon travel mode has no significant effect on travel mode choice, while the "behavior attitude" of high-carbon mode has a significant negative effect on it. The performance of the model is tested by monthly income sensitivity analysis, it shows that when the income increases, low-carbon travel proportion decreases gradually while the mid-carbon and high-carbon travel proportions rise, and the increasing proportion of high-carbon travel decreases when considering the low-carbon psychological factors.