[Paper Review] Mobsimilarity: Vector Graph Optimization for Mobility Tableau Comparison

Yuhao Yao, Haoran Zhang+5 more|arXiv (Cornell University)|Apr 27, 2021

Human Mobility and Location-Based Analysis39 references4 citations

TL;DR

This paper introduces Mobsimilarity, a vector graph optimization method for comparing mobility tableaus—high-dimensional representations of population flows between location pairs. By minimizing spatial transformation cost between vector graphs, it captures volume, spatial, structural, and inclusiveness similarities, outperforming traditional OD matrix methods in mobility estimation validation and cross-city comparison case studies.

ABSTRACT

Human mobility similarity comparison plays a critical role in mobility estimation/prediction model evaluation, mobility clustering and mobility matching, which exerts an enormous impact on improving urban mobility, accessibility, and reliability. By expanding origin-destination matrix, we propose a concept named mobility tableau, which is an aggregated tableau representation to the population flow distributed between different location pairs of a study site and can be represented by a vector graph. Compared with traditional OD matrix-based mobility comparison, mobility tableau comparison provides high-dimensional similarity information, including volume similarity, spatial similarity, mass inclusiveness and structure similarity. A novel mobility tableaus similarity measurement method is proposed by optimizing the least spatial cost of transforming the vector graph for one mobility tableau into the other and is optimized to be efficient. The robustness of the measure is supported through several sensitive analysis on GPS based mobility tableau. The better performance of the approach compared with traditional mobility comparison methods in two case studies demonstrate the practicality and superiority, while one study is estimated mobility tableaus validation and the other is different cities' mobility tableaus comparison.

Motivation & Objective

• To address the limitations of traditional OD matrix-based mobility comparison, which often overlooks structural and spatial nuances in population flow data.
• To develop a comprehensive similarity measure that captures volume, spatial distribution, mass inclusiveness, and structural patterns in mobility data.
• To enable more accurate evaluation of mobility models and clustering by leveraging aggregated, high-dimensional representations of human movement.
• To provide a robust, efficient framework for comparing mobility patterns across different cities or time periods using vector graph optimization.

Proposed method

• Introduces the concept of a 'mobility tableau' as a vector graph representation of population flows between origin-destination pairs in a study area.
• Models mobility tableaus as vector graphs where nodes represent locations and edges represent flow volumes, enabling multi-dimensional similarity analysis.
• Proposes a transformation cost minimization framework that optimizes the least spatial cost to map one vector graph to another, using graph alignment techniques.
• Employs a vectorized optimization approach to efficiently compute similarity by minimizing displacement cost while preserving structural and volumetric fidelity.
• Incorporates sensitivity analysis to validate robustness under GPS data noise and sparsity.
• Applies the method to real-world mobility datasets, comparing results against baseline OD matrix methods.

Experimental results

Research questions

• RQ1 How can mobility data be represented in a way that captures not only flow volumes but also spatial distribution and structural patterns?
• RQ2 To what extent does the proposed vector graph-based similarity measure outperform traditional OD matrix comparisons in mobility model evaluation?
• RQ3 How robust is the Mobsimilarity method under noisy or sparse GPS data conditions?
• RQ4 Can the method effectively compare mobility patterns across different cities with varying urban structures?
• RQ5 What are the relative contributions of volume, spatial, structural, and inclusiveness similarities in mobility tableau comparison?

Key findings

• The Mobsimilarity method achieves superior performance in mobility model validation compared to traditional OD matrix-based approaches, with improved accuracy in similarity detection.
• The method demonstrates robustness under GPS data noise and sparsity, maintaining consistent similarity measurements across varying data quality levels.
• In cross-city mobility comparison, Mobsimilarity successfully captures structural and spatial differences between urban mobility patterns, revealing meaningful distinctions.
• The inclusion of mass inclusiveness and structure similarity significantly enhances the discriminative power of the similarity measure beyond volume-only comparisons.
• The optimization framework is computationally efficient, enabling scalable comparison of large-scale mobility tableaus.
• Sensitivity analysis confirms that the method is stable and reliable across diverse real-world mobility datasets.