A statistical framework for analyzing activity pattern from GPS data
By: Haoyang Wu, Yen-Chi Chen, Adrian Dobra
Potential Business Impact:
Tracks where people go to understand their habits.
We introduce a novel statistical framework for analyzing the GPS data of a single individual. Our approach models daily GPS observations as noisy measurements of an underlying random trajectory, enabling the definition of meaningful concepts such as the average GPS density function. We propose estimators for this density function and establish their asymptotic properties. To study human activity patterns using GPS data, we develop a simple movement model based on mixture models for generating random trajectories. Building on this framework, we introduce several analytical tools to explore activity spaces and mobility patterns. We demonstrate the effectiveness of our approach through applications to both simulated and real-world GPS data, uncovering insightful mobility trends.
Similar Papers
Graph Network Modeling Techniques for Visualizing Human Mobility Patterns
Social and Information Networks
Maps how people move to predict city changes.
Learning Universal Human Mobility Patterns with a Foundation Model for Cross-domain Data Fusion
Machine Learning (CS)
Helps cities plan roads and traffic better.
Modeling Human Spatial Mobility Patterns with the Lévy Flight Cluster Model
Methodology
Tracks people's movements to create fake, private maps.