A Simple Walk Model for Reproducing Power Laws in Human mobility

TL;DR

This paper introduces a simple walking model that reproduces key power-law patterns in human mobility data, offering a new perspective on the origins of these statistical laws.

Contribution

A novel minimalistic walking model that naturally generates observed power laws in human mobility without predefining rules, unlike previous models.

Findings

The model reproduces power laws in jump size and waiting time distributions.

It generates visitation frequency distributions consistent with Zipf's law.

The model explains the emergence of scaling laws in human movement patterns.

Abstract

Identifying statistical patterns characterizing human trajectories is crucial for public health, traffic engineering, city planning, and epidemic modeling. Recent developments in global positioning systems and mobile phone networks have enabled the collection of substantial information on human movement. Analyses of these data have revealed various power laws in the temporal and spatial statistical patterns of human mobility. For example, jump size and waiting time distributions follow power laws. Zipf's law was also established for the frequency of visits to each location and rank. Relationship $S(t)\sim t^\mu$ exists between time t and the number of sites visited up to that time t. Recently, a universal law of visitation for human mobility was established. Specifically, the number of people per unit area $\rho(r,f)$, who reside at distance r from a particular location and visit that location f times in a given period, is inversely proportional to the square of rf, i.e., $\rho(r,f) \propto (rf)^{-2}$ holds. The exploration and preferential return (EPR) model and its improved versions have been proposed to reproduce the above scaling laws. However, some rules that follow the power law are preinstalled in the EPR model. We propose a simple walking model to generate movements toward and away from a target via a single mechanism by relaxing the concept of approaching a target. Our model can reproduce the abovementioned power laws and some of the rules used in the EPR model are generated. These results provide a new perspective on why or how the scaling laws observed in human mobility behavior arise.