PROGRAM: the TRANSITDATA2019 abstracts and presentations will be soon available on the website.
9h00-10h30: SESSION I: DEMAND ESTIMATION – chair: Mark HICKMAN, University of Queensland
- A common challenge with automated public transport data is that passenger movements are not explicitly or comprehensively described within these data. This session explores emerging methods to infer public transport demand from these data. The methods explored in this session include techniques to estimate passenger destinations, vehicle loads, origin-destination flows, and complete journeys on public transport. Moreover, the session highlights applications of these methods to large-scale transit systems.
- Generating network-wide travel diaries using smartcard data (15) – Oded Cats (TU Delft), Alex Vermeulen, Matej Cebecauer, Erik Jenelius and Yusak Susilo.
- Comparing origin-destination matrices derived from smart card data with a large scale OD survey. Results from a case study in Lyon (13) – Oscar Egu (Keolis Lyon) and Patrick Bonnel (LAET – ENTPE)
- Inferring trip destinations in transit smart card data using a probabilistic topic model (18) – Zhanhong Cheng (McGill University), Martin Trépanier (Polytechnique Montréal and CIRRELT) and Lijun Sun (McGill University and CIRRELT)
- Real-time estimation of bus passenger OD patterns based on AVL data (10) – Wenzhe Sun (Kyoto University), Jan-Dirk Schmöcker (Kyoto University) and Koji Fukuda (Hitachi Kyoto University Laboratory)
10h30-11h00: coffee break
11h00-12h30: SESSION II: DEMAND CHARACTERISTICS AND PATTERNS – chair: Martin TREPANIER, Polytechnique Montréal and CIRRELT
- Public transport data may come from many sources. Starting from this, the presentations of this session show how data mining methods will help to characterise and group not only transport user behaviour, but also other objects such as stations and land use around them.
- The unexpected flexibility of public transit usage revealed from mining Israel’s smartcard data (37) – Itzhak Benenson (Tel-Aviv University) and Eran Ben-Elia (Ben-Gurion University of the Negev)
- Mining regional passenger mobility activity using multiple data sources: A case study in Paris area (45) – Yao Zhou, Xiaoyan Xie and Fabien Leurent (LVMT, UMR-T 9403, Ecole des Ponts, IFSTTAR, UPEM)
- A density-based smart card data classification algorithm applied to land use analysis and infrastructure performance measure (51) – Li He (Polytechnique Montréal), Martin Trépanier (Polytechnique Montréal), Bruno Agard (Polytechnique Montréal), Marcela Munizaga (Universidad de Chile) and Benjamin Bustos (Universidad de Chile)
- Investigating non-rail usage at large railway stations in city centres (46) – Napat Kittiwongsophon and Taku Fujiyama (University College London)
14h00-15h30: SESSION III: FLOW PREDICTIONS – chair: Taku FUJIYAMA, University College London
- This session will include studies on transport demand and flows. Increasingly available data enables fresh approaches. The session will cover studies on ridership fluctuation, bus passenger flow (including crowding levels) and road traffic flow. We will see a variety of data types, including roadside CCTV, smartcard systems, trip planner usage, weather data, operation logs, land use and Point of Interest data.
- Forecasting bus ridership with trip planner usage data: a machine learning application (44) – Jop van Roosmalen, Chintan Amrit, Engin Topan and Niels van Oort (TU Delft)
- Prediction of bus passenger flow using Deep Learning (36) – Walid Kheriji, Sami Kraiem, Guilhem Sanmarty, Ghazaleh Khodabandelou and Fouad Hadj Selem (Vedecom)
- Impact of weather, service disruptions and other events on transportation demand (29) – Simon Lepage and Catherine Morency (Polytechnique Montréal, CIRRELT)
- Roadway Traffic Flow Estimation using Video Imagery Data Collected from Transit Bus Cameras (58) – Rabi Mishalani, Mark Mccord, Benjamin Coifman and Giovani Hansel (The Ohio State University)
16h00-17h30: SESSION IV: TRAVEL TIME PREDICTIONS – chair: Nigel WILSON, MIT
- Travel time is one of the key variables affecting service quality on public transport, as reflected both in concerns about the waiting time for service and the likelihood of arriving at the destination on time. This session deals with three aspects of travel time prediction with the use of automated data sources, all of which can contribute to better passenger information and operations plans,, and eventually help in improving the performance of many public transport systems. These three presentations deal with different aspects of the integrated public transport system, but make use of the most important automated data sources: AFC, AVL and load weigh systems.
- Transit Travel Time Distributions Estimation Based on Passive AFC Data (1) – Amr M. Wahaballa (Aswan University), Fumitaka Kurauchi (Gifu University), Jan-Dirk Schmöcker (Kyoto University) and Takenori Iwamoto (Shizuoka Railway Co.)
- Investigating the effects of in-car passenger volume and its distribution on train dwell time (30) – Konstantina Argyropoulou (University College London), Howard Wong and Taku Fujiyama
- Real-Time Bus Travel Time Prediction: Exploring Machine Learning and External Factors (20) – Ryan Williams, Amer Shalaby and Siva Srikukenthiran (University of Toronto)
At-stop bus arrival interval prediction based on analysis of travel time forecasting error (34) – Agostino Nuzzolo and Antonio Comi (University of Rome Tor Vergata)
8h30-9h00: Welcome coffee
9h00-10h30: SESSION V: ROUTE CHOICE – chair: Jan-Dirk SCHMOECKER, Kyoto University
- Predicting route choice of passengers is the core of any transit assignment approach. Smart card and other automated data allow us though in most cases no direct estimation of the routes taken since often only tap-on and tap-off are recorded (or only one of the two). Furthermore, in rail bound networks the tap-on and tap-off points are mostly not at the vehicle but at gates so that waiting time and travel times can not be distinguished but must be inferred. Finally, even if these problems are solved and interchange points would also be recorded, the effect of crowding and capacity constraints on route choice require a modelling approach to infer their importance. This session includes four papers that address these problems to advance our understanding of route choice estimation through automated data.
- Understanding passenger path choice in congested metro networks – the case of reverse routing (38) – Morten Eltved (Technical University of Denmark), Haris Koutsopoulos (Northeastern University), Nigel Wilson (Massachusetts Institute of Technology) and Kerem Tuncel (Northeastern University)
- Estimating Passengers’ Path Choice Using Automated Data in Urban Rail Systems (55) – Zhenliang Ma (Monash University and Northeastern University), Haris Koutsopoulos (Northeastern University), Yiwen Zhu (Microsoft Bay Area) and Yunqing Chen (Monash University)
- Accounting for the consideration of common lines in route choice models of public transport passengers (24) – Jacqueline Arriagada (Universidad de Chile), Marcela Munizaga (Universidad de Chile), Angelo Guevara (Universidad de Chile) and Carlo Prato (The University of Queensland)
- Decomposing journey time variance on urban metro systems via semiparametric mixed methods (16) – Ramandeep Singh, Daniel J. Graham and Richard J. Anderson
11h00-12h30: SESSION VI: NETWORKS ASSESSMENT – chair: Rabi MISHALANI, The Ohio State University
- This session focuses on assessing transit networks from a variety of perspectives. In the first presentation, origin-destination matrices inferred from multi-day smart card data are used to evaluate a new public transport network aimed to reduce user cost while maintaining current operating costs. In the second, a new type of topological representation of transit networks that incorporates wait, transfer, and in-vehicle times derived from timetables is proposed and applied to several networks for a comparative analysis. In the third, an in-depth analysis and characterisation of travel time reliability is proposed and applied where the effects of travel time reliability on travellers’ mode choice decisions are analysed. In the fourth presentation, smart card and AVL data are used to study the impact of a new metro line on travel patterns, travel times, and reliability as experienced by passengers.
Evaluating the Robustness of a New Proposed Public Transport Network Using Multiday Origin-Destination Matrix Inferred from Smart Cards (3) – Renato O. Arbex (University of São Paulo)
- Enhanced complex network representation of public transport for accessibility assessment based on General Transit Feed Specification data (12) – Ding Luo (Delft University of Technology), Oded Cats and Hans van Lint
- Public transport travel time reliability across modes and space (41) – Jaime Soza-Parra (Pontifical Catholic University of Chile), Sebastián Raveau and Juan Carlos Munoz
- Impact analysis of a new metro line in Amsterdam using automated data sources (35) – Malvika Dixit, Ties Brands, Oded Cats, Niels van Oort and Serge Hoogendoorn (Delft University of Technology)
14h00-15h30: SESSION VII: RAIL OPERATIONS AND DISRUPTIONS – chair: Niels VAN OORT, TU Delft
Evaluating train timetable robustness using historical train traffic records (6) – Yasufumi Ochiai (Chiba Institute of Technology), Yuta Shibata and Norio Tomii
- Representation Learning of public transport data. Application to event detection (40) – Kevin Pasini (Université Paris-Est IFSTTAR and IRT SystemX), Allou Samé (Université Paris-Est), Mostepha Khouadjia (IRT SystemX), Fabrice Ganansia (SNCF- Innovation & Recherche) and Latifa Oukhellou (Université Paris-Est IFSTTAR)
- Using smart card data to analyse the disruption impact on urban metro systems (48) – Nan Zhang, Daniel J. Graham, Jose M. Carbo and Daniel Hörcher (Imperial College London)
- Predicting and clustering station vulnerability in urban networks (4) – Menno Yap and Oded Cats (Delft University of Technology)
16h00-17h30: SESSION VIII: TRAVEL INFORMATION – chair: Marcela MUNIZAGA, Universidad de Chile
- The availability of online data about the operation of public transport systems, provides a wonderful opportunity to improve the travel experience through personalized travel information and recommendations. This session will include three papers: the first one focuses on customer segmentation with the aim of providing targeted information, the second one will show us the results of a simulation analysis to study the effects of providing online crowding information on traveler’s behaviour, and the third paper will present us a trip planner that combines long term and short term information to improve the recommendations made to public transport users.
- Data-Driven Customer Segmentation and Personalized Information Provision in Public Transit (59) – Abhishek Arunasis Basu, Jinhua Zhao, Haris Koutsopoulos and Rabi Mishalani (The Ohio State University)
- Investigating the effects of real-time crowding information (RTCI) systems in urban public transport under different demand conditions (39) – Arkadiusz Drabicki (Cracow University of Technology), Rafał Kucharski (Cracow University of Technology) and Oded Cats (Delft University of Technology)
- Predictive Multimodal Trip Planner: A New Generation of Urban Routing Services (49) – Ahmed Amrani (IRT SystemX), Kevin Pasini (IRT SystemX and Université Paris‐Est IFSTTAR) and Mostepha Redouane Khouadjia (IRT SystemX)
8h30-9h00: Welcome coffee
9h00-10h30: SESSION IX: PUBLIC TRANSPORT RIDERSHIP – chair: Fumitaka KURAUCHI, Gifu University
- This session focuses on how we can forecast transit ridership as well as how to evaluate congestion in transit system. The first paper proposes to apply a modern data science method inferring OD patterns using smartcard data, and the second paper discusses about the behaviour of infrequent passengers using smartcard data. Then, the third and fourth papers attempts to evaluate the level of service of transit system both from viewpoints of passenger and train congestion, using macroscopic fundamental diagram and fail-to-board probability, respectively.
- Extraction of public transportation networks from Openstreetmap data and estimation of OD patterns through a graph convolutional approach (57) – Kouji Fukuda (Hitachi Kyoto University Laboratory), Saeed Maadi (University of Glasgow), Kai Shen (University of Glasgow) and Jan Dirk Schmoecker (Kyoto University)
- Infrequent Public Transport Use: An Investigation with Smart Card Data (60) – Mark Hickman and Tianwei Yin (The University of Queensland)
- Emperical investigation of fundamental diagram for urban rail transit using tokyo’s commuter rail data (54) – Daisuke Fukuda (Tokyo Institute of Technology), Masahiro Imaoka and Toru Seo
- A Probabilistic Traffic Model to Estimate Fail-to-Board Probabilities in Transit Lines on the basis of AFC and AVL data (25) – Fabien Leurent (LVMT Ecole des ponts ParisTech – ENPC) and Thomas Jasmin
10h30-11h00: Welcome coffee
11h00-12h30: SESSION X: TRANSIT FARES & PERFORMANCE – chair: Amer SHALABY, University of Toronto
- This session will feature presentations on the impacts of fare payment policies and technologies on system revenue, ridership and service reliability. The presentation will demonstrate the application of data mining and machine learning methods to the analysis of public transport sensor data.
- Evaluating the impact of fare capping and guaranteed best fare policies using smart card validations data and machine learning (42) – Alfred Ka Kee Chu (Autorité régionale de transport métropolitain, Montréal), André Lomone (Autorité régionale de transport métropolitain, Montréal) and Robert Chapleau (École Polytechnique de Montréal)
How do payers and evaders travel on public transport? (23) – Angel Cantillo (Pontifical Catholic University of Chile) and Sebastián Raveau
- Analyzing the Effect of Mandatory Smart-Card Use and Priority Lanes on Public Transport Reliability (56) – Nir Ron and Yuval Hadas (Bar-Ilan University)
Bus drivers with different driving behaviors impact the performance of public transport systems (14) – Yerly Martinez (Pontifical Catholic University of Chile), Juan Carlos Muñoz, Felipe Delgado and Kari Watkins
14h00-15h30: SESSION XI: DEMAND ANALYTICS – chair: Patrick BONNEL, Head of Transport Department, ENTPE
- The rapid evolution of new mobility services such as Ride-hailing, car sharing, bike sharing… impact transit use. Session will discuss these impacts through different analysis and modelling methods
- Understanding the Ridership Decline on a Disaggregated Spatial and Temporal Level in Five Cities (5) – Simon Berrebi (Georgia Tech), Sanskruti Joshi, Taylor Gibbs and Kari Watkins
- Understanding the Influence of the Built Environment and Zonal Attributes on the Relationship between Public Transit and Ride Hailing Services (19) – Patrick Loa, Sanjana Hossain and Khandker Habib (University of Toronto)
- Longitudinal modeling of the daily subway ridership in Montreal: what is the influence of alternative modes of transport? (47) – Elodie Deschaintres (Polytechnique Montréal), Catherine Morency (Mobility Chair, CIRRELT / Polytechnique Montréal) and Martin Trepanier (Mobility Chair, CIRRELT / Polytechnique Montréal)
- Exploring the Ridership Impacts of Ride-hailing Service on a Multimodal Transit System across Time Periods (26) – Wenting Li, Amer Shalaby (University of Toronto) and Khandker Nurul Habib
15h30-16h00: coffee break
16h00-17h30: SESSION XII: VISUALIZATION & APPLICATIONS – chair: Brendon HEMILY, Public Transportation Consultant
- The availability of new sources of passive data that can be combined with existing data sources, and the development of new methodologies and AI-based tools is expanding the horizons for public transportation planning analyses. These developments are enabling the processing of large volumes of data, new approaches to data mining and data fusion, as well as the development of visualization tools for conducting spatial analysis and creating dashboards for planning and executive management. This session will illustrate examples of these developments and how it is leading to a better understanding of customer experience and/or more sophisticated insights to assist planning and management.
- Analysis and visualization of the French transportation network based on massive passive data (31) – Paul Soulé (SNCF) and Théo Delecour
- Visualization tools for spatio-temporal time-series analysis with context awareness: Montreal subway case (21) – Florian Toqué (Université Paris-Est, IFSTTAR, COSYS-GRETTIA), Etienne Côme (Université Paris-Est, IFSTTAR, COSYS-GRETTIA), Latifa Oukhellou (Universite Paris-Est, IFSTTAR, COSYS-GRETTIA) and Martin Trépanier (Ecole Polytechnique de Montréal)
- A data-intensive software application for customer-centered transit planning (27) – Charles Fleurent, Timothy Spurr, Flavie Gagnon-Pontbriand and Loïc Bodart (GIRO Inc.)
- Measure Door-to-Door Mobility Through GPS Tracks & Big Data Methodologies (43) – Sylvain Coppéré, Stéphane Mastalerz, Félix Motot and Marwann Ghanem (Kisio Etudes & Conseil)