We recently utilized spatiotemporal point data of commercial facilities, such as shown in Figure 1. We have a list of commercial facilities from 1995 to 2020 in Shibuya-ku, Tokyo. Geocoding of the addresses generates spatial point data, which serve as a basis for quantitative spatial analysis in a GIS (Geographic Information System).

One application of the data is to analyze the appearance and disappearance of points over time. The study aims to develop measures for evaluating the frequencies of appearances and disappearances. One possible measure that can serve this purpose is the ratio of events to points calculated in each lattice cell. This measure, however, cannot distinguish between the disappearance of 1 out of 2 points and that of 1000 out of 2000 points. Whereas the former can easily happen by chance, the latter implies the occurrence of a major problem. To resolve this, we developed a new method for analyzing the appearance and disappearance of points. The method is based on the likelihood of appearances and disappearances under a random situation, thus avoiding the problem mentioned above.

We applied the method to the analysis of the openings and closures of commercial facilities in Shibuya-ku. Figures 2-4 show the distributions of the proposed measures calculated for clothing stores, restaurants, and clinics, respectively.

Purple: Frequent openings and closures
Light green: Infrequent openings and closures
Orange: Frequent openings and infrequent closures
Blue: Infrequent openings and frequent closures

Openings and closures are both frequent around Shibuya station in all the figures. Interestingly, openings and closures are infrequent in the immediate vicinity of Shibuya station in Figure 3. This is because many old and traditional restaurants in this area have long-term support from loyal customers. Existing methods cannot provide these findings, thus supporting the effectiveness of the proposed method.

The above is one example of the applications of spatiotemporal point data of commercial facilities. Spatiotemporal point data have rapidly become available due to the development of new data acquisition tools. Analysis of such data, on the other hand, has not yet been discussed extensively. Methods for analyzing spatiotemporal data need to be developed to reveal the underlying structure of spatiotemporal point patterns.

〈LINKS〉

https://www.tandfonline.com/doi/abs/10.1080/13658816.2018.1538517
https://www.sciencedirect.com/science/article/pii/S0198971518305829

Text:Yukio Sadahiro (Professor)
English proofreading:David Buist (Senior project specialist)