ESTIMATION THE NUMBER OF TOURISTS BASED ON MESH POPULATION WITH GPS DATA

Abstract

The increase in tourism has led to the notion of “over-tourism” and furthermore causes dis-satisfaction of travellers with their tourism experience. Not least due to the economic significance of tourism addressing these issues is important. A large number of tourism studies are based on survey analysis. On the other hand, an increasing number of studies are utilizing various “big data” sets in recent years. In particular mobile phone data and GPS tracking data are becoming available. Considering this, we estimate the number of tourists using GPS traces in different areas of Kyoto city by regression analysis and hierarchical linear models (HLM). The GPS data are limited and we supplement these with access cost information. The data are verified with aggregate spatial information, “mesh population”, obtained from a mobile phone operator. We obtain satisfactory results from the regression analysis and further improved results from HLM. We show that with at least 4500 GPS traces the estimated tourist variance reduces significantly. We find that GPS traces correspond to mesh population well when we consider the person who visit tourist areas more than 0.3 per day as tourists. We suggest this provides a guideline for studies aiming to use GPS traces for population estimation. We also show that the GPS data can capture monthly differences in tourism patterns. We also find that other data than GPS data like travel cost or POIs can adjust the differences between distributions of mesh population and GPS data.