Estimation of walking patterns in a touristic area with Wi-Fi packet sensors

We propose a methodology for using the records from Wi-Fi packet sensors to model route choice and time spent at different locations within a tourist area. In contrast to other route choice problems, in tourist areas it is often the case that "the route is the goal," so we expect route choice behavior to be less destination oriented. Data were collected from Higashiyama Ward, Kyoto, Japan, which is highly frequented by tourists. We utilized Wi-Fi packet sensors, which avoid the need for mobile applications but provide only spot data. Based on the data, we first discuss the extraction of tourist trip chains and the construction of a simplified time-space network with "stay" and "move" links. The network and link attributes are based on the location of the sensors as well as map data. Recursive Logit (RL) models are then employed to formulate tourists' route choice behavior. The existence of different kinds of "points of interests" (POIs) is shown to explain route choice as well as the time spent on roads with shops, restaurants and sightseeing spots.