Discrete Choice Models

We identify new solutions to transportation problems, on the ground, in the air, or on the sea, transport of people or goods, whatever the mode. We focus on technical solutions, but also on their impact on the system as a whole. We are also interested in the interactions of the transportation systems with the land use, the economy, the environment, etc.

Migration and Discrete Choice Models (MIGDCM)

International migration is at the forefront of policy debates in most countries around the world. In industrialized nations, the proportion of foreigners in total population increased from 4.5 to 12 percent between 1960 and 2019, stirring up fears about economic costs for natives, loss of national identity, and integration issues. In poor countries, international migration raises concerns about the brain drain of highly-skilled workers, as college and university graduates have a much greater propensity to emigrate internationally than the less educated. Hence, the questions of how many people migrate (i.e., migration intensity}, which people migrate first or are more likely to migrate (i.e., migrants' selection), and where migrants choose to settle (i.e., migrants' sorting) have been analyzed from all possible angles in recent literature. Specifically, understanding how people revise their decisions about whether to emigrate, and where to, when facing changes in the global environment is of crucial importance for decision-makers.

Principal investigator

Michel Bierlaire

Project managers

Silvia Varotto, Evangelos Paschalidis

Sponsor

Swiss National Science Foundation

Period

March 01, 2022-March 31, 2025

External collaboration

Prof. Michel Beine (University of Luxemburg)

LaTeX description

Optimization of individual mobility plans to simulate future travel in Switzerland

This project, joint with Swiss Federal Railways (SBB) will develop a new activity-based modelling approach based on optimization of individual daily mobility plans. This approach will be implemented within SBB's existing nationwide model for Switzerland for investment and service planning decisions for future transportation.

Principal investigator

Michel Bierlaire

Project manager

Tim Hillel

Sponsor

Innosuisse (Swiss Innovation Agency)

Period

September 01, 2020-March 01, 2022

Collaborators

Janody Pougala, Rico Krueger

External collaboration

Patrick Manser

External collaboration

Wolfgang Scherr

LaTeX description

Activity scheduling an rhythmic style: multi-day modeling of mobility habits

A growing body of research shows that the traditional approaches employed to plan and forecast travel behaviors are not equipped to deal with the heterogeneity of behaviors over time, space, and social spheres. Therefore, travel policies struggle to reconcile social inclusivity, sustainability and network efficiency. Two ENAC laboratories propose to join forces, in mathematical modeling and quantitative sociology, to develop a novel multi-day activity-scheduling framework to forecast travel demand. Integrating day-to-day correlations in travel behaviors will lead to a better understanding of the motives behind travel decisions, and will unveil more facets of individual decision making for better predictions of their daily mobility choices. This research uses the MOBIS dataset, a 8-week / 3700-respondent travel survey conducted in Switzerland in 2019. The forecasting model will be applied to the Swiss synthetic population and several scenarios will be considered.

Principal investigator

Janody Pougala

Project manager

Janody Pougala

Sponsor

School of Architecture, Civil and Environmental Engineering (ENAC), École polytechnique fédérale de Lausanne

Period

October 01, 2020-October 01, 2021

Collaborator

Marija Kukic

External collaboration

Marc-Edouard Schultheiss

LaTeX description

Incorporating competitors' behavior in demand based optimization models

The starting point of this research project is the general framework developed in the SNF research project 200021_165636 entitled “Incorporating advanced behavioral models in mixed integer linear optimization”. The project aims at formulating explicitly the complex interaction between the supply and demand actors. We hypothesize that the decisions of supply actors (such as capacity, assortment, and price) are highly influenced by the decisions of other actors, through competition within the market. In this project, we will therefore explicitly consider multiple operators that compete for the same pool of customers. Each operator will take the supply-side decisions that optimize its performance function (e.g., maximization of revenue). Non-cooperative game theory is commonly used to model such oligopolistic competition. We plan to investigate how to integrate these games in the already developed framework. In this context, the objective will be to analyze the concept of equilibrium, that is, stationary states of the system where no actor has an incentive to change his/her decisions.

Principal investigator

Michel Bierlaire

Project manager

Virginie Lurkin

Sponsor

Swiss National Science Foundation

Period

November 01, 2017-October 31, 2020

Collaborator

Stefano Bortolomiol

LaTeX description

Activity based travel demand forecasting

This research project aims to update and improve the microscopic activity-based demand model developed and maintained by SBB. Specifically the research intends to address the following questions: 1. Ownership of mobility instruments: Which metrics and specifications can be added to the current model, in order to improve its ability to forecast mid-and long-term ownership of mobility instruments? More specifically, how can the notion of accessibility be integrated to the current model to capture more complex mode interactions? 2. Mode choice model: Can a tour-based approach be used to model mode choice? In addition, how can the processes to estimate destination and mode choice (currently nested) be combined to generate results that are consistent with observed mobility behaviors at different time horizons (short, mid, and long-term)? 3. Tour and activity generation: How can the generation of tours and activity patterns be combined to allow modelling of joint decisions?

Principal investigator

Michel Bierlaire

Project manager

Tim Hillel

Sponsor

Swiss Federal Railways (SBB)

Period

March 01, 2019-March 01, 2020

Collaborator

Janody Pougala

LaTeX description

Incorporating advanced behavioral models in mixed integer linear optimization

Discrete choice models are used for detailed representation of the "demand". However, their complexity makes mathematical formulations highly non convex in the explanatory variables. On the other hand, Mixed Integer Linear Programs (MILP) are optimization problems with discrete variables that are used in many applications to design and configure the "supply". In this project, we propose a new modeling framework that allows to include any random utility model in a mixed integer optimization formulation. The heterogeneity of demand is captured within the general choice model framework and the offers are tailored in a way that is beneficial for users as wells as providers. The main objective of the project is to obtain a framework that is both general, so that it can be applied in many applications, and operational, so that it can be used in practice.

Principal investigator

Michel Bierlaire

Project manager

Shadi Sharif Azadeh

Sponsor

Swiss National Science Foundation

Period

April 01, 2016-March 31, 2019

Collaborator

Meritxell Pacheco

LaTeX description

TRANS-FORM: Smart transfers through unravelling urban form and travel flow dynamics

TRANS-FORM, a cooperation between universities, industrial partners, public authorities and private operators, will develop, implement and test a data driven decision making tool that will support smart planning, and proactive and adaptive operations. The objective of the project is to better understand transferring dynamics in multi-modal public transport systems and develop insights, strategies and methods to support decision makers in transforming public transport usage to a seamless travel experience by using smart data. The tool will integrate new concepts and methods of behavioral modelling, passenger flow forecasting and network state predictions into real-time operations. TRANS-FORM is funded under the European Commission Horizon 2020 ERA-NET program.

Principal investigator

Michel Bierlaire

Project managers

Shadi Sharif Azadeh, Riccardo Scarinci, Yuki Oyama

Sponsor

The Federal Department of the Environment, Transport, Energy and Communications (DETEC)

Period

March 03, 2016-March 03, 2019

Collaborators

Nicholas Molyneaux, Nikola Obrenovic

LaTeX description

Drive For You: a driving assistant tool to detecting pedestrians

This project aims to develop an onboard pedestrian tracking system to assist the driver detect them and, ultimately, to increase security. The project, with a duration of four years, will focus on pedestrians detection, tracking and trajectory prediction, and will be will be closely related to vehicle command strategies. The project is part of the automated driving research Chair "Drive for you" led by MINES ParisTech in partnership with French industrialists and three prestigious academic institutions the Ecole Polytechnique Fédérale de Lausanne EPFL (Switzerland), the University of Shanghai Jiao Tong (China) and the University of Berkeley (USA). Supported by the Foundation MINES ParisTech, with Valeo industrial, PSA Peugeot Citroën and Safran contributing 3.7 million euros in funds, the Chair will work for five years on the subject of automated driving. The three main objectives are expand knowledge of self-driving vehicles, develop intelligent onboard systems, get self-driving vehicles on the road in Asia, Europe and the United States.

Principal investigator

Michel Bierlaire

Project manager

Riccardo Scarinci

Sponsor

MINES ParisTech

Period

December 01, 2014-November 30, 2018

External collaboration

Damien Matti (EPFL Signal Processing Laboratory 5 LTS5)

LaTeX description

Electric vehicle adoption dynamics: exploring market potentials

This project proposes innovative methods to identify the determinants of acceptance of alternative vehicles and their impact on everyday mobility. The greatest challenge faced by the promoters of the transition towards this low carbon engine technology lies in understanding how consumers accept the financial and lifestyle investments associated with the leap from traditional to electric powertrains. This project proposes innovative methods to identify the determinants of acceptance of alternative vehicles and their impact on everyday mobility. A deeper understanding of adoption dynamics is critical to predict who will opt for EVs when and under which conditions. This project will focus on innovative data-collection and modelling methodologies to uncover the acceptance of EVs at different stages of market-penetration (considering inexperienced/experienced users, early pioneers/late-adopters). A thorough analysis of the consumer decision-making process will lead to uncovering the barriers and success factors related to EV uptake and to forecast buying and usage behaviours related to new vehicle classes.

Principal investigator

Michel Bierlaire

Project manager

Matthieu de Lapparent

Sponsor

Nissan

Period

April 01, 2014-March 31, 2017

Collaborator

Anna Fernandez Antolin

External collaboration

Dr. Amanda Stathopoulos, Northwestern University

LaTeX description

Revisiting the Route Choice Problem: A Multi-Level Modeling Framework for Route Choice Analysis Based On Latent Constructs

This project aims at developing behavioral models integrating the latent constructs that drive travel behavior. The objective of the project is to set the foundations of an innovative framework that facilitates the analysis and prediction of route choice behavior. Motivated by the complexity of the route choice models, we build on solid ground of the current state of the art and add on it by proposing a new approach that reduces model complexity and brings great flexibility to the analysis. The approach is inspired by the fact that people break down the complexity of the environment by forming representations of their surrounding space. Following this reasoning, we replace the conventional representation of routes which is based on paths, i.e. link-by-link sequences on the network model, with more aggregate elements that we denote as mental representation items. This key feature of the framework allows us to reduce the complexity of the model and at the same time is more behaviorally realistic.

Principal investigator

Michel Bierlaire

Project manager

Matthieu de Lapparent

Sponsor

Swiss National Science Foundation

Period

October 01, 2013-September 30, 2016

Collaborator

Evanthia Kazagli

External collaboration

Dr. Amanda Stathopoulos, Northwestern University

LaTeX description

ACTUM

The ACTUM research collaboration project involves development of an activity-based modelling framework for Copenhagen that is aimed at handling the future requirements for transport policy evaluations in Denmark.

Principal investigator

Michel Bierlaire

Project manager

Matthieu de Lapparent

Sponsor

The Danish Council for Strategic Research

Period

March 01, 2011-March 01, 2016

More information:

http://www.dtu.dk/subsites/ACTUM/English.aspx

LaTeX description

Route choice models and smart phone data

Smart phones are able to collect a great wealth of information about the mobility patterns of individuals, including their itinerary. In addition to GPS tracks, smart phones contain several sensors and software revealing various aspects of the user's behavior (nearby Bluetooth devices, WiFi antenna, media players, etc.) Route choice models rely on the knowledge of the origin and the destination of the trip, and the transportation mode. They may also benefit from other contextual data, such as the trip purpose or the chain of subsequent activities. We plan to investigate how to infer this information from the smart phone data.

Principal investigator

Michel Bierlaire

Project manager

Jingmin Chen

Sponsor

Swiss National Science Foundation

Period

October 01, 2010-September 30, 2013

Collaborator

Jingmin Chen

LaTeX description

A DYNAMIC VEHICLE CHOICE FORECASTING FRAMEWORK

The aim of this research project is to develop a dynamic model framework to forecast car demand. The model-system jointly accounts for behavioural variables (e.g. household evolution, individual motivations), institutional factors (e.g. incentives and taxation) and market features (e.g. new fuel or vehicle technology launches) to forecast vehicle acquisition. The project benefits from a large-scale, longitudinal dataset on vehicle transactions from major European markets. Methodologically the framework consists of a system of aggregate, disaggregate and duration-based choice models, accounting for dynamic features of the decision at each stage. Importantly, the proposed dynamic framework enhances realism by representing the durability of possession, rapid changes in value across product and ownership cycles and uncertainty about future market developments and institutional factors. The proposed framework sheds light on the timing and type of vehicle ownership, illustrating the benefits of taking the first steps away from a static approach. The findings will aid the prediction of consumer behaviour in response to critical policy variables, such as incentives, alongside market dynamics, such as the appearance and diffusion of novel technologies. A deeper understanding of future vehicle choice behaviour is critical given sizable institutional and industrial investments to promote alternative vehicle technologies with a final aim to improve energy efficiency in the transportation sector. More information http://transport.epfl.ch/car-market-structure

Principal investigator

Michel Bierlaire

Project manager

Amanda Stathopoulos

Sponsor

PSA Peugeot-Citroën

Period

April 01, 2012-April 30, 2013

Collaborator

Aurélie Glerum

More information:

http://transport.epfl.ch/car-market-structure

LaTeX description

OPTIMA: Inferring transport mode preferences from attitudes

Swiss Post and PostBus have launched a research project to optimize mobility in cooperation with the Federal Institute of Technology in Lausanne (EPFL), represented by EPFL Transportation Center, under Professor Michel Bierlaire. This will show the market potential for combined mobility. Among other things it will look at the factors that influence travellers in their choice of transport. At a later stage, target group-specific new transport offerings and services will be developed and tested, which will have a lasting influence on this choice. The three years research project is conducted by Professors Vincent Kaufmann (LASUR) and Michel Bierlaire (TRANSP-OR).

Principal investigator

Michel Bierlaire

Project manager

Anne Curchod

Sponsor

LaPoste

Period

February 01, 2009-June 30, 2012

Collaborators

Aurélie Glerum, Bilge Atasoy, Antonin Danalet

External collaboration

Vincent Kaufman (LASUR)

External collaboration

Prisca Faure (CEAT)

External collaboration

Martin Schuler (CEAT)

LaTeX description

Demande de consommateurs pour des véhicules électriques

Evaluation de la demande des consommateurs pour des véhicules électriques en vue du lancement d'une telle gamme par Renault

Principal investigator

Michel Bierlaire

Project manager

Michaël Thémans

Sponsor

Renault Suisse SA

Period

August 01, 2010-April 30, 2011

Collaborator

Aurélie Glerum

LaTeX description

Behavioural modelling of human experts for scene analysis

Modelling the global appraisal process of experts in the context of scene analysis.

Principal investigator

Michel Bierlaire

Project managers

Javier Cruz, Thomas Robin

Sponsor

Swiss National Science Foundation

Period

October 01, 2007-September 30, 2010

LaTeX description

Disaggregate behavioral models exploiting data from Nokia devices

The projects aims at investigating how to perform an appropriate fusion of data from smartphones in order to predict the mobility behavior of a given individual.

Principal investigator

Michel Bierlaire

Project manager

Jeffrey Newman

Sponsor

Nokia Research Center

Period

December 01, 2008-June 30, 2010

Collaborator

Jingmin Chen

LaTeX description

Path to purchase

To understand consumer behavior in his or her path to purchase.

Principal investigator

Michel Bierlaire

Project manager

Kamran Houshang Pour

Sponsor

Nestlé

Period

July 01, 2006-June 30, 2008

Collaborators

Thomas Robin, Gianluca Antonini

External collaboration

Jean-Philippe Thiran, ITS, EPFL

LaTeX description

Development of route choice models

Design new discrete choice models capturing route choice behavior

Report(s):

Bierlaire, M., and Frejinger, E. Route choice modeling with network-free data

Frejinger, E., and Bierlaire, M. Sampling of Alternatives for Route Choice Modeling

Frejinger, E., and Bierlaire, M. Capturing correlation in large-scale route choice models

Principal investigator

Michel Bierlaire

Project manager

Emma Frejinger

Sponsor

Swiss National Science Foundation

Period

May 01, 2005-April 30, 2008

LaTeX description

Einbezug von Reisekosten bei der Modelliering des Mobilitätsverhalten

Development of route choice models from RP and SP data

Principal investigator

Michel Bierlaire

Project manager

Emma Frejinger

Sponsor

Swiss Federal Road Office

Period

November 01, 2005-February 28, 2007

Collaborator

Jelena Stojanovic

External collaboration

Prof. K. W. Axhausen (IVT, ETHZ)

External collaboration

Prof. R. Maggi (Instituto Ricerche Economiche, Univesità della Svizzera Italiana)

LaTeX description

Behavioral model-based scene analysis

Development of disaggregate behavior model for scene analysis in the context of video surveillance

Principal investigator

Michel Bierlaire

Project manager

Michel Bierlaire

Sponsor

NCCR Interactive Multimodal Information Management

Period

July 01, 2002-June 30, 2005

Collaborator

Gianluca Antonini

External collaboration

Jean-Philippe Thiran, LTS, EPFL

LaTeX description

Développement de modèles suisses de prédiction de la demande en transport pour des applications en temps réel (Development of Swiss models for transportation demand prediction in the context of traffic management systems)

Development of operational disaggregate demand models for efficient dynamic traffic managment systems

Principal investigator

Michel Bierlaire

Project manager

Michel Bierlaire

Sponsor

VSS - Association of Swiss Road and Traffic Engineers

Period

January 01, 2003-December 31, 2004

Collaborator

Michaël Thémans

External collaboration

Prof. K. W. Axhausen, IVT, ETHZ

External collaboration

A. Robert-Grandpierre, bureau d'ingénieurs RGR-SA

External collaboration

P. Widmer, Büro Widmer, Frauenfeld

LaTeX description

SIMBAD: Simulation-Based Automatic Detection

Develop a model-based detection system where the scene analysis is supported by a crowd simulator

Principal investigator

Michel Bierlaire

Project manager

Mats Weber

Sponsor

Commission pour la technologie et de l'innovation, Office Fédéral de la formation professionnelle et de la technologie.

Period

October 01, 2002-September 30, 2004

Collaborator

Gianluca Antonini

External collaboration

Mats Weber, ROSO, EPFL

External collaboration

Visiowavw S.A

External collaboration

Jean-Philippe Thiran, LTS, EPFL

LaTeX description

Large-scale discrete choice models

We investigate methodological developments designed to handle discrete choice models with a large number of alternatives

Report(s):

Bierlaire, M., Bolduc, D., and McFadden, D. The estimation of Generalized Extreme Value models from choice-based samples

Bierlaire, M., and Lucadamo, A. Sampling of alternatives using multidimensional analysis

Principal investigator

Michel Bierlaire

Project manager

Michel Bierlaire

Sponsor

Transport and Mobility Laboratory, EPFL

Period

March 25, 2003-March 25, 2004

LaTeX description