Mobility Patterns: Integrating App-Based Data and Advanced Clustering for Enhanced Understanding of Human Mobility

This paper presents the findings of a comprehensive study aimed at simplifying human mobility analysis by emphasizing main destinations. The research combines traditional survey methods with innovative app-based data collection to capture a detailed picture of mobility. The study analyzes human mobility by applying clustering algorithms such as DBSCAN and KMeans, coupled with precise mobility data collected via the "Mobility Report" app from 284 participants over an extended period (17 months). A key innovation of our study lies in developing a method to identify, compare, and cluster individuals based on their most visited locations. The analysis reveals that focusing on frequently visited destinations effectively represents most mobility behaviors. Clustering algorithms categorize users based on visitation patterns, showing that clustering based on distance metrics does not outperform frequency-based clustering. Results indicate that a few key destinations capture the majority of mobility behaviors, with distance being not correlated with visitation frequency. These findings suggest future surveys shifting towards identifying the most frequented locations rather than detailing daily itineraries. This approach offers a refined understanding of urban mobility, potentially informing better infrastructure development and transportation planning. The study’s methodology and findings contribute to advancing urban mobility research, highlighting the repetitive nature of human movement and the importance of main destinations in mobility analysis.