基于低碳心理潜变量Logit模型的出行方式预测模型

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

为深入解析低碳出行行为机理,提出了将低碳心理变量引入离散选择模型的研究方法。基于计划行为理论考虑低碳出行心理因素,建立了多原因多指标潜变量模型。将潜变量模型预测后的潜变量代入多项Logit模型,构建了带低碳出行心理变量的混合选择模型。以镇江市居民通勤调查样本为研究对象,实证结果表明:相对传统离散选择模型,所建的混合选择模型预测精度整体提高了2.45%;低碳心理变量对于出行方式选择的影响各不相同,低碳出行方式"行为态度"对出行方式选择没有显著性影响,而高碳出行方式"行为态度"对低碳出行有显著性负影响。通过对月收入的敏感性分析检验模型性能,结果显示:随着收入的增长,低碳出行方式占比逐渐降低,其他两种方式占比变化相反,在考虑低碳心理因素后,高碳出行方式占比上升幅度降低。

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.