Impacts of Seasonal Factors on Travel Behavior: Basic Analysis of GPS Trajectory Data for 8 Months

Abstract

Understanding people’s activities and travel behaviors has gained attention in service research field as well as in transportation research field. Recently, there are a lot of studies utilizing GPS (Global Positioning System) trajectory data to analyze travel behaviors after identifying each trip. Although transportation service level (e.g., travel time or waiting time) and our travel behaviors would change due to weather and seasonal factors, there is no research to evaluate an automated detection/identification model for GPS trajectory data. In this study, we compare a trip frequency (trip purpose of shopping and health, which are nonmandatory trips) and an accuracy of the detection/identification model by using long-term person trip survey data, which is conducted for each 4 months in summer and winter in Hakodate city, Japan. From the results, we confirm that a variation of car trip frequency is small and a tendency of mode change from bicycle in summer to walking in winter is strong due to snowy roadside condition. Moreover, random forest method as the detection/identification model has small effect to seasonal variations if multi-seasonal data is combined.