Characterization of cities based on world grid square statistics about specific properties

This article proposes how to capture characteristics of cities based on world grid square statistics about the number of properties. First, how to compute grid square statistics from point data is explained, and a definition of the world grid square coding system is introduced. Second, a method to characterize cities based on grid square statistics about the number of properties is proposed. Empirical analysis for seven selected types of properties (automated teller machines (ATM), hostels, toilets, banks, post offices, hotels, and cafes) in Kyoto, Japan, and Erevan, Armenia, is conducted. Both the ordinary least squares (OLS) and reduced major axis (RMA) regression analyses for the linear relationship among the numbers of properties in terms of world grid square are used with the t-test in order to measure co-occurrence among properties. Moreover, a method to compare two cities based on matrices of regression coefficients is proposed. The comparative analysis between the OLS regression analysis and the RMA regression analysis is conducted. We discuss a relationship between the proposed method and statistical disclosure control (SDC) methods introduced in official statistics and an application of the proposed method to private information regarding confidentiality of private information with geographical positions.