A Method to Measure the Degree of the Favorite Location Visiting of Mobile Objects

Abstract

To understand the mobility of humans or things, it is necessary to measure the degrees of location visits in everyday mobility. In this paper, we discuss measures that can present human preferences to certain locations based on location data and analysis. From raw positioning data and the concept of location clusters, which are sets of positioning data representing location areas, several measures can be deduced. First, the location point and location area can be separated because visiting a pin point location is different from visiting a certain area. Second, the number of visits to a location and the duration of a visit to a location have different meanings. Third, the rank of the location visited is sometimes more meaningful than the absolute counts. In consideration of these aspects, we established six basic measures and two derived measures. The actual calculation of each measure requires raw positioning data to be processed. The raw positioning data were collected by volunteers over several years of their everyday lives. All measures for multiple volunteers were generated and analyzed for verification. The processing of raw positioning data to generate measures requires a vast number of calculations, like big data processing. As a solution, we implemented a generation process using the programming language R; GPGPU technology was utilized to derive numerical results within areas on able time limit with considerable speed-ups, because an undesirably large amount of time was required to process measures with CPU-only machines.