A Statistical Method for Identifying Areas of High Mobility Applied to Commuting Data for the Country of New Zealand

TL;DR

This paper introduces a statistical approach using Markov chains to identify high-mobility locations in commuting data, demonstrated through a case study of New Zealand.

Contribution

It proposes a novel permutation-based statistical method to detect mobility loci in human movement data, advancing mobility analysis techniques.

Findings

Identified key mobility loci in New Zealand commuting data.

Validated method against known commuting patterns.

Showed the method's effectiveness in highlighting high-mobility areas.

Abstract

Human mobility describes physical patterns of movement of people within a spatial system. Many of these patterns, including daily commuting, are cyclic and quantifiable. These patterns capture physical phenomena tied to processes studied in epidemiology, and other social, behavioral, and economic sciences. This paper advances human mobility research by proposing a statistical method for identifying locations that individual move to and through at a rate proportionally higher than other locations, using commuting data for the country of New Zealand as a case study. These locations are termed mobility loci and they capture a global property of communities in which people commute. The method makes use of a directed-graph representation where vertices correspond to locations and traffic between locations correspond to edge weights. Following a normalization, the graph can be regarded as a Markov chain whose stationary distribution can be calculated. The proposed permutation procedure is then applied to determine which stationary distributions are larger than what would be expected, given the structure of the directed graph and traffic between locations. The results of this method are evaluated, including a comparison to what is already known about commuting patterns in the area as well as a comparison with similar features.