Inferring human mobility using communication patterns

TL;DR

This paper demonstrates that human mobility patterns can be effectively inferred from aggregated mobile phone call data alone, leveraging call frequency and geographical distance, thus avoiding privacy issues associated with direct location tracking.

Contribution

The study introduces a simple model that predicts mobility using only call data and distance, highlighting the social dimension of movement and privacy preservation.

Findings

Mobility can be predicted from call frequency and distance.

The model incorporates social aspects of human movement.

Aggregated call data suffices for mobility inference.

Abstract

Understanding the patterns of mobility of individuals is crucial for a number of reasons, from city planning to disaster management. There are two common ways of quantifying the amount of travel between locations: by direct observations that often involve privacy issues, e.g., tracking mobile phone locations, or by estimations from models. Typically, such models build on accurate knowledge of the population size at each location. However, when this information is not readily available, their applicability is rather limited. As mobile phones are ubiquitous, our aim is to investigate if mobility patterns can be inferred from aggregated mobile phone call data alone. Using data released by Orange for Ivory Coast, we show that human mobility is well predicted by a simple model based on the frequency of mobile phone calls between two locations and their geographical distance. We argue that the strength of the model comes from directly incorporating the social dimension of mobility. Furthermore, as only aggregated call data is required, the model helps to avoid potential privacy problems.