摘要：AbstractThis paper presents an empirical assessment of the spatial transferability of tour-based time-of-day choice models across different counties in the San Francisco Bay Area of California. Model transferability was assessed using two different approaches: (1) estimation-based approach, and (2) application-based approach. The former approach allows the analyst to test which specific coefficients in the model are transferable and which are not, while the latter approach tests the transferability of a model as a whole. In addition, the hypothesis that pooling data from different geographical contexts helps in estimating better transferable models than those developed from a single context was tested. The estimation-based approach yields encouraging results in favour of model transferability, with a majority of the coefficients in a pooled model found to be transferrable. Pooling data from different geographical contexts appears to help in developing better transferable models but careful attention is needed in selecting the geographical contexts to pool the data from.
关键词：Time-of-day choice;transferability;multinomial logit model;pooling data
