How Short Is Long Enough? Modeling Temporal Aspects of Human Mobility Behavior Using Mobile Phone Data

Area of Science:

• Human Mobility and Behavior Science
• Geospatial Data Analysis
• Urban Planning and Public Health

Background:

• Location-sensing technologies generate valuable time-location data for understanding human mobility.
• Current human activity studies often overlook the crucial aspect of data collection duration.
• The temporal dimension of human mobility behavior remains underexplored in existing literature.

Purpose of the Study:

• To determine the minimum number of days required to capture organized human activity episodes.
• To investigate how this minimum duration varies across individuals with diverse demographic and socioeconomic characteristics.
• To address the gap in understanding the temporal requirements for human mobility data collection.

Main Methods:

• Examined Kullback-Leibler divergence index distributions to establish a minimum observation period.
• Employed Bayesian profile regression models to analyze variations in minimal observation days across subgroups.
• Utilized time-location data from location-sensing technologies for human mobility behavior analysis.

Main Results:

• The estimated minimum number of days needed to capture routine activity patterns is 13.5 days (SD = 6.64).
• Participant demographics (age, household size), employment status, and accessibility to amenities (downtown, food, physical activity) significantly influence mobility behavior.
• The economic status of the residential environment is also a key factor affecting temporal mobility patterns.

Conclusions:

• A minimum of approximately 13.5 days of data is necessary to reliably capture routine human mobility patterns.
• Individual heterogeneity in demographic and socioeconomic factors necessitates tailored data collection periods for accurate mobility behavior analysis.
• Future research should consider these temporal and individual factors when designing studies on human mobility.