基于Mixed Logit模型的路线选择行为量化分析

Quantitative analysis of traveler route choice behavior based on Mixed Logit model

为了探讨路况信息动态变化对出行路线选择行为的影响程度,采用正交试验法设计出行意愿调查问卷,针对有驾车经验的出行者进行调查。假定时间参数和费用参数为常数、费用参数服从正态分布和对数正态分布,分别建立了出行路线选择的Logit模型和Mixed Logit模型,并对解释变量影响路线改变倾向进行了量化分析。结果表明,Mixed Logit模型的最大似然值、优度比和命中率都高于Logit模型。出行者对行程时间和时间信息动态变化的敏感程度十分接近,敏感度最高的是出行费用。个人属性中的年龄、驾龄和收入是影响路线选择行为的最显著因素。

To explore the effect of the urban road traffic dynamics on the traveler route choice behavior, a stated preference survey questionnaire was designed using the orthogonal design method. The survey was conducted with the travelers with driving experience in Dalian city. With the time parameter and the cost parameter fixed as well as the cost parameter obeying the normal distribution and lognormal distribution, a Logit model and a Mixed Logit model were built for the route choice analysis, and a quantitative analysis was performed for the route change trend caused by variables. The model estimation results indicated that the maximum likelihood, the values of goodness-of-fit index, and hit ratio of the Mixed Logit model are better than those of the Logit model. The traveler sensitivity for the travel time and dynamic change of time information are similar, and the most sensitive parameter is the travel cost. The traveler age, years of driving experience and income are the most significant factors affecting the route choice behavior.