Identifying the Multimodal Hierarchy of Public Transit Systems Using Itinerary Data

Basic Information

• Paper ID: 2509.24220
• Title: Identifying the Multimodal Hierarchy of Public Transit Systems Using Itinerary Data
• Authors: Junhee Lee¹, Seungmo Kang²,, Jinwoo Lee²,
• Classification: cs.CE (Computational Engineering, Finance, and Science)
• Publication Status: Currently under review
• Paper Link: https://arxiv.org/abs/2509.24220

Abstract

As urban transportation systems integrate traditional and emerging transit modes, public transit systems have become increasingly complex. Complementary or competitive relationships between different transit modes influence users' multimodal travel path choices. This paper introduces the concept of a macroscopic multimodal hierarchy framework, wherein travel follows an "ascending-descending" pattern: beginning and ending with low-level transit modes that provide high accessibility (such as walking), while utilizing efficient high-level transit modes (such as subway) in the middle segments. The authors propose a method for identifying urban multimodal transit hierarchies based on multimodal smart card itinerary data and validate it using real data from Seoul and surrounding metropolitan areas in South Korea.

Research Background and Motivation

Problem Definition

With urbanization and transportation technology development, modern urban transit systems exhibit complex multimodal and multilevel characteristics. Traditional transportation planning typically constructs transit hierarchies based on planners' design concepts but lacks in-depth analysis of actual passenger usage patterns.

Research Significance

1. Theoretical Value: Understanding the hierarchical structure of multimodal transit systems is crucial for optimizing transportation network design
2. Practical Implications: Can provide decision support for transportation planners based on actual passenger behavior
3. Policy Guidance: Facilitates the formulation of more effective multimodal transit integration strategies

Limitations of Existing Methods

1. Spatial Homogeneity Assumption: Previous studies (e.g., Daganzo and Ouyang, 2019) assumed spatial homogeneity, overlooking the role of local transit modes
2. Lack of Empirical Validation: Discrepancies exist between theoretically designed hierarchies and actual passenger usage patterns
3. Limited Method Applicability: Existing methods struggle to accurately capture the hierarchical status of light rail, community buses, and other local transit modes

Research Motivation

This paper aims to identify multimodal transit hierarchies based on actual passenger travel behavior by analyzing real smart card data, bridging the gap between theoretical design and actual usage.

Core Contributions

1. Proposed a Novel Hierarchy Identification Method: Developed a multimodal transit hierarchy identification method based on the "ascending-descending" theoretical framework without requiring spatial homogeneity assumptions
2. Constructed a Quantitative Analysis Framework: Defined quantitative metrics such as transfer rate matrices and hierarchical distances to measure hierarchical relationships between transit modes
3. Validated Method Effectiveness: Verified the practical utility of the method using over 10 million real itinerary records from the Seoul metropolitan area
4. Revealed Regional Variations: Discovered significant differences in the hierarchical status of the same transit mode across different regions

Methodology Details

Task Definition

Input: Multimodal smart card itinerary dataset containing transfer information for M transit modes Output: Hierarchical ranking values H*ᵢ ∈ 0,1 for each transit mode, where higher values indicate higher hierarchical status Constraints: Exclude pure walking trips; assume the first and last segments of each trip are walking

Model Architecture

2.1 Single-Region Hierarchy Analysis

Stage Segmentation:

• Divide ascending and descending stages based on the midpoint of total trip distance Lₙ
• Ascending stage: transfers from origin to Lₙ/2
• Descending stage: transfers from Lₙ/2 to destination

Transfer Rate Matrix Construction:

Ascending stage transfer rate: Aᵢⱼ = aᵢⱼ/(aᵢⱼ + aⱼᵢ)
Descending stage transfer rate: Dᵢⱼ = dᵢⱼ/(dᵢⱼ + dⱼᵢ)

Hierarchical Distance Calculation:

Ascending stage hierarchical distance: A*ᵢⱼ = Aᵢⱼ - Aⱼᵢ ∈ [-1,1]
Descending stage hierarchical distance: D*ᵢⱼ = Dⱼᵢ - Dᵢⱼ ∈ [-1,1]

Average Hierarchical Ranking:

A*ᵢ = 1/2(∑(A*ᵢⱼ/(M-1))) + 1/2
D*ᵢ = 1/2(∑(D*ᵢⱼ/(M-1))) + 1/2
H*ᵢ = (A*ᵢ + D*ᵢ)/2

2.2 Multi-Region Hierarchy Analysis

For systems containing multiple regions P:

• Inter-Regional Analysis: Calculate H*ᵢ^pq using cross-regional trip data N^pq
• Intra-Regional Analysis: Calculate H*ᵢ^pp using within-region trip data N^pp

Technical Innovations

1. No Spatial Homogeneity Assumption: Allows different transit modes to have different hierarchical statuses across regions
2. Bidirectional Hierarchy Verification: Validates hierarchy stability through consistency between ascending and descending stages
3. Quantitative Assessment Framework: Provides comparable numerical hierarchical ranking metrics

Experimental Setup

Dataset

• Data Source: T-money transit card data from Seoul metropolitan area, November 12, 2019
• Coverage: Seoul, Gyeonggi Province, Incheon
• Data Scale: 10,264,700 qualified trip records
• Data Distribution:
  • Intra-Seoul trips: 7,500,141
  • Seoul to Gyeonggi/Incheon: 1,365,156
  • Gyeonggi/Incheon to Seoul: 1,399,403

Transit Mode Classification

The study encompasses 6 transit modes (M=6):

1. Walking (i=1)
2. Community Bus (i=2)
3. City Bus (i=3)
4. Intercity Bus (i=4)
5. Light Rail (i=5)
6. Subway (i=6)

Regional Division

• Region 1: Seoul
• Region 2: Gyeonggi Province and Incheon

Experimental Results

Main Results

Intra-Seoul Hierarchy (N₁₁)

Based on analysis of 7.5 million intra-Seoul trips, the following hierarchical ranking was obtained:

1. Subway: Highest hierarchy (H*≈0.8)
2. Light Rail: High hierarchy (H*≈0.7)
3. City Bus: Medium hierarchy (H*≈0.6)
4. Community Bus: Low hierarchy (H*≈0.4)
5. Walking: Lowest hierarchy (H*≈0.2)

Key Finding: Hierarchical rankings in ascending and descending stages show high consistency (Aᵢ ≈ Dᵢ ≈ H*ᵢ), validating the correctness of the "ascending-descending" theory.

Inter-Regional Hierarchy (N₁₂ and N₂₁)

Hierarchical ranking with intercity buses included:

1. Intercity Bus: Highest hierarchy
2. Subway: Second-highest hierarchy
3. Light Rail: Upper-middle hierarchy
4. City Bus: Medium hierarchy
5. Community Bus: Lowest hierarchy

Regional Variation Analysis

The study reveals significant regional variations:

1. Subway vs. Intercity Bus: Subway has relatively higher hierarchical status in Seoul due to high station density; intercity buses are more prevalent in Gyeonggi Province and Incheon
2. City Bus: Relatively lower hierarchical status in Gyeonggi Province and Incheon
3. Relative Hierarchy Changes: The inclusion of intercity buses causes relative hierarchical status changes for other transit modes

Experimental Findings

1. Hierarchy Consistency: The same transit mode exhibits consistent hierarchical status in both ascending and descending stages
2. Regional Adaptability: Transit mode hierarchical status varies according to regional characteristics
3. Relativity Principle: Hierarchy structure is relative; introduction of new transit modes affects overall ranking

Related Work

Theoretical Foundation

• Daganzo and Ouyang (2019): Proposed the low→high→low multimodal travel theoretical framework
• Wang et al. (2020): Emphasized the importance of understanding transit hierarchy for effective planning
• Oh et al. (2024): Validated the hierarchical model of ascending-descending multimodal itinerary formation

Advantages of This Paper

1. Method Innovation: No spatial homogeneity assumption required; broader applicability
2. Data-Driven: Based on large-scale real data rather than theoretical assumptions
3. Quantitative Analysis: Provides comparable numerical evaluation framework

Conclusions and Discussion

Main Conclusions

1. Theory Validation: Empirical data supports the "ascending-descending" multimodal travel theory
2. Hierarchy Identification: Successfully identified transit mode hierarchical ranking in Seoul metropolitan area
3. Regional Variations: The same transit mode exhibits different hierarchical status across regions

Limitations

1. Data Limitations: Uses only single-day data, which may introduce temporal bias
2. Geographic Constraints: Method applicability in other cities requires further verification
3. Transit Mode Coverage: Does not include all emerging transit modes (e.g., bike-sharing, ride-hailing services)

Future Directions

1. Time Series Analysis: Study temporal evolution patterns of hierarchy structure
2. Multi-City Comparison: Verify method generalizability across different cities
3. Emerging Transit Modes: Incorporate hierarchical analysis of more novel transit modes

In-Depth Evaluation

Strengths

1. Theoretical Contribution: Proposes hierarchy identification method based on actual passenger behavior, bridging theory-practice gap
2. Method Innovation: Overcomes spatial homogeneity assumption limitations, improving method applicability
3. Data Scale: Uses over 10 million real data records, ensuring result reliability
4. Quantitative Framework: Establishes complete mathematical model and computational procedures

Weaknesses

1. Missing Temporal Dimension: Analyzes only single-day data, lacking time series analysis
2. Insufficient Factor Analysis: Lacks in-depth exploration of key factors influencing hierarchy structure
3. Limited Method Validation: Lacks comparative experiments with other hierarchy identification methods

Impact

1. Academic Value: Provides new research paradigm for transit system hierarchy analysis
2. Practical Value: Directly applicable to urban transportation planning and policy formulation
3. Reproducibility: Clear method description with good reproducibility

Application Scenarios

1. Urban Transportation Planning: Provides data support for multimodal transit network design
2. Policy Formulation: Assists in formulating transit integration and optimization strategies
3. Academic Research: Provides analytical tools for transportation system complexity research

References

1. Daganzo, C.F. and Ouyang, Y., 2019. Public transportation systems: Principles of system design, operations planning and real-time control.
2. Oh, J., Jung, Y., Lee, C. and Lee, J., 2024. A Hierarchical Model for Ascending-Descending Multimodal Itinerary Formation.
3. Wang, Z., Luo, D., Cats, O. and Verma, T., 2020. Unraveling the hierarchy of public transport networks.

Overall Assessment: This paper proposes an innovative and practical multimodal transit hierarchy identification method with significant value in both theoretical and practical dimensions. Despite certain limitations, its core contributions establish new research directions in the transportation systems analysis field.