出行方式与出发时间联合选择的分层Logit模型

基于随机效用最大化理论,选取出行者特征、行程特性与出行方式服务水平作为效用变量,以出行方式与出发时间作为选择肢,构建了出发时间位于下层与出行方式位于下层的2种居民出行NL模型。分析了北京市居民出行样本数据,并模拟了在早高峰时段对小汽车出行收取费用时,小汽车出行者出行行为的变化。计算结果表明:与传统MNL模型相比,NL模型具有更好的统计学特征,调整后的拟合优度由0.338增大至0.404;在2种NL模型中,出发时间位于下层的结构对样本数据的适应性更强;当早高峰时段小汽车出行收取费用为5元时,72.6%的小汽车出行者坚持原有出行方式与出发时间,22.4%的小汽车出行者坚持小汽车方式,但会改变出发时间,4.8%的小汽车出行者改用公共交通方式,但出发时间不变,仅0.2%的小汽车出行者同时改变出行方式与出发时间;当收取费用为10元时,51.7%的小汽车出行者坚持原有出行方式与出发时间,40.4%的小汽车出行者坚持小汽车方式,但会改变出发时间,7.9%的小汽车出行者改用公共交通方式,但出发时间不变;当收取费用为20元时,27.5%的小汽车出行者坚持原有出行方式与出发时间,60.6%的小汽车出行者坚持小汽车方式,但会改变出发时间,11.9%的小汽车出行者改用公共交通方式,但出发时间不变。

Nested logit model of combined selection for travel mode and departure time

Based on the maximum random utility theory,traveler characteristic,travel characteristic and the service level of travel mode were taken as utility variables,travel mode and departure time were taken as alternative parts,and two nested logit(NL) models were built,one structure with departure time located in lower layer and another structure with travel mode located in lower layer.The sample data of resident travel in Beijing City were analyzed,and the travel behavior changes of car travelers were simulated when different car travel costs were charged in morning peak period.Calculation result shows that compared with traditional MNL model,there is better statistic characteristic in NL model.After adjustment,the goodness of fit increases from 0.338 to 0.404.In the two NL models,the structure with departure time located in lower layer has stronger adaptability on sample data than the structure with travel mode located in lower layer.While car travel cost in morning peak period is 5 yuan,72.6% of car travelers will still insist on original travel mode and departure time,22.4% of car travelers will still insist on original travel mode,but will change departure times,4.8% of car travelers will turn to public transit,but will still insist on original departure time,and only 0.2% of car travelers will change travel mode and departure time simultaneously.While car travel cost in morning peak period is 10 yuan,51.7% of car travelers will still insist on original travel mode and departure time,40.4% of car traveler will still insist on original travel mode,but will change departure times,and only 7.9% of car travelers will turn to public transit,but will still insist on original departure time.While car travel cost in morning peak period is 20 yuan,27.5% of car travelers will still insist on original travel mode and departure time,60.6% of car travelers will still insist on original travel mode,but will change departure times,and only 11.9% of car travelers will turn to public transit,but will still insist on original departure time.6 tabs,3 figs,22 refs.