Characterizing the activity patterns of outdoor jogging using massive multi-aspect trajectory data

Zongshun Tian,Wei Yang,Tong Zhang,Tinghua Ai,Yang Wang

DOI: https://doi.org/10.1016/j.compenvurbsys.2022.101804

2022-01-01

Abstract:Characterizing activity pattern such as behavior preferences or habits is crucial in many fields. However, existing studies mainly focus on the spatial-temporal dimensions of raw trajectory, but ignore the context information in multi-aspect trajectory that affects behavior significantly. In this paper, we present a data-driven framework to characterize outdoor jogging activity patterns with massive multi-aspect trajectory data. In our framework, a novel multi-aspect trajectory Latent Dirichlet Allocation (MAT-LDA) model is presented to discover latent activity patterns from multi-aspect trajectories. Specifically, the model inherits from LDA, but extends its topics and words to mine the combined patterns in multi-aspects. Then, clustering analysis is performed to find and characterize the jogger groups with similar preference patterns. Experiments with real jogging GPS tracks recorded by 16,643 users' fitness app show that the MAT-LDA model can efficiently discover the latent activity patterns and quantify the correlations and interdependencies between patterns of multi-aspect attributes. Moreover, many interpretable preferences are discovered at individual level, and jogger groups (e.g., mini groups, jog hobbyist) with common context-aware preferences are revealed to understand fitness jogging. Our method can capture activity behavior preferences of multiple aspects from multi-aspect trajectory data, and our work can enrich outdoor fitness application with interpretable preference patterns.