Investigating Cyclic Visit Pattern of Mobility Through Analysis of Geopositioning Data

Abstract

Intuitions guide us that there are cyclic patterns for a person to visit a location, and there is a tendency of multiple cycles in visiting patterns. Nowadays, it is possible for a person to collect personal mobility data due to the help of smartphones and other portable devices. These devices collects raw geolocation (or geopositioning) data and the set of geolocation data can be analyzed in various ways. Based on location clusters distilled from raw geolocation data, we can establish mobility model of a person and investigate cyclic patterns of a person to visit location clusters. Based on the aggregate personal mobility models collected over several years, we calculated and analyzed the cluster revisiting time and visualized it as a graph. Regarding geolocation data for location clusters as set of time sequence, number of visiting cluster is measured in a unit of minutes. The number of visits from whole data is normalized in every 15 min. For various geolocation data set of a volunteer, cyclic patterns of a visit are examined in terms of autocorrelation, autocovariance and intervisiting time.