抄録
In Japan, a national interregional travel demand survey has been conducted since 1990, but it has some limitations related to measuring seasonal demand variation as well as collecting trip frequency. A new survey was proposed and carried out in 2015 and early 2016 to capture travel demand and trip frequency in one year. To explore new survey data, in this study, a model, using eigenvector spatial filters specification, is developed to estimate interregional travel demand across the four seasons. As a result, the significance is seen in all likelihood ratio tests in all considered models, which proves that the spatial filter models are preferable to the unfiltered ones. The explanatory variables in each season have significantly different effects on travel demand, which would not appear in the model with cross-sectional data. For further studies, some remaining issues should be solved such as adding time effect or integration of more impedance variables into the model, developing specific models for each purpose or each travel mode, or an overall model considering a seasonal dummy variable.
