Characterizing People's Daily Activity Patterns in the Urban Environment: A Mobility Network Approach with Geographic Context-Aware Twitter Data

TL;DR

This study introduces a geographic context-aware mobility network approach using Twitter data to analyze detailed daily activity patterns in urban environments, revealing common location and activity motifs in Chicago and Boston.

Contribution

It presents a novel method combining geo-located Twitter data with land use information to identify and analyze granular urban activity motifs.

Findings

Identified 16 location-based motifs covering over 83% of networks.

Dissected location motifs into 16 activity-based motifs explaining 57% of behaviors.

Revealed unique activity motifs embedded in complex urban activities.

Abstract

People's daily activities in the urban environment are complex and vary by individuals. Existing studies using mobile phone data revealed distinct and recurrent transitional activity patterns, known as mobility motifs, in people's daily lives. However, the limitation in using only a few inferred activity types hinders our ability to examine general patterns in detail. We proposed a mobility network approach with geographic context-aware Twitter data to investigate granular daily activity patterns in the urban environment. We first utilized publicly accessible geo-located tweets to track the movements of individuals in two major U.S. cities: Chicago and Greater Boston, where each recorded location is associated with its closest land use parcel to enrich its geographic context. A direct mobility network represents the daily location history of the selected active users, where the nodes are physical places with semantically labeled activity types, and the edges represent the transitions. Analyzing the isomorphic structure of the mobility networks uncovered 16 types of location-based motifs, which describe over 83% of the networks in both cities and are comparable to those from previous studies. With detailed and semantically labeled transitions between every two activities, we further dissected the general location-based motifs into activity-based motifs, where 16 common activity-based motifs describe more than 57% transitional behaviors in the daily activities in the two cities. The integration of geographic context from the synthesis of geo-located Twitter data with land use parcels enables us to reveal unique activity motifs that form the fundamental elements embedded in complex urban activities.