Estimating origin-destination matrices in Helsinki's public transport through multi-source data fusion

Thumbnail Image

Files

master_Fernandes_Fermin_Eduardo_2025.pdf (26.75 MB)

URL

Journal Title

Journal ISSN

Volume Title

School of Science | Master's thesis
Unless otherwise stated, all rights belong to the author. You may download, display and print this publication for Your own personal use. Commercial use is prohibited.

Authors

Date

2025-09-28

Department

Major/Subject

Machine Learning, Data Science and Artificial Intelligence

Mcode

Degree programme

Master's Programme in Computer, Communication and Information Sciences

Language

en

Pages

72

Series

Abstract

Accurate Origin-Destination (OD) matrices are essential for public transport planning, yet challenging to obtain in Proof-of-Payment systems without entry-exit records. This thesis presents a multi-source data fusion framework to estimate network-level OD matrix for Helsinki's bus network by integrating trajectories from HSL's mobile application (TravelSense) with Automated Passenger Count (APC) stop totals. Using Iterative Proportional Fitting (IPF), we balance sparse TravelSense seed matrices with APC marginals at route level and aggregate to a network-level OD matrix. A streaming pipeline enables the processing and fusion of the large-scale source datasets, and an interactive web visualization supports exploration of edge loads, stop net flows, and OD arcs. In a weekday-morning case study, route-level IPF achieves good marginal fit, showing logical OD patterns, while network aggregation recovers high trip coverage on major corridors. External validation against Telia CrowdInsights shows limited cell-wise agreement at fine granularity due to scope and spatial aggregation effects, but metropolitan scale patterns are consistent and expected to agree more under coarser aggregation and transit-likelihood filtering. Overall, the approach yields balanced route-level matrices that align with observed APC marginals and a network-wide OD that covers the bulk of scheduled service, with visible logical patterns. Results demonstrate that fusing emerging trajectory data with traditional counts yields operationally useful OD products in Proof-of-Payment (POP) systems and provides a practical foundation for routine OD estimation.

Description

Supervisor

Saramäki, Jari

Thesis advisor

Huang, Zhiren

Keywords

origin-destination matrices, public transport, data fusion, smartphone-based mobility data, travel behavior, Automated Passenger Counting (APC), Iterative Proportional Fitting (IPF), Helsinki region

Other note

Citation

Permanent link to this item

https://urn.fi/URN:NBN:fi:aalto-202510218417

Collections

[dipl] Perustieteiden korkeakoulu / SCI