A SPARC Short Course on
Advanced Choice Modelling Methods in an Evolving
Urban Travel Behaviour Landscape

3rd - 7th February 2020

Venue: CCE, IISc

Background: Effective management of urban transportation systems requires a thorough understanding of the choices and behaviour of the users of the system. The emergence of new mobility modes and technologies – mobility as a service (MaaS), electric vehicles (EVs), and novel public transit systems that integrate conventional transit with on-demand mobility modes – makes it more important to understand and model the travel choices people make in complex urban settings. A widely used methodological paradigm to understand transportation user behaviour is econometric choice modelling. Disaggregate choice models are at the heart of travel demand model systems that are used to forecast urban mobility patterns and system performance under alternative scenarios of population growth, land-use, transportation system characteristics, technology solutions, and policy strategies. In addition, choice models are employed to analyse user choices in a variety of fields, including marketing, environmental economics, geography, urban planning, and tourism. Examples include, but are not limited to, users’ valuation of non-market goods (e.g., air quality) in environmental economics, households’ residential location and employers’ firm location choices in geography and urban planning, and vacation destination and mode choices in tourism. In short, choice modelling is an important technique for analysing disaggregate, agent-level behaviour and for forecasting system-level outcomes relevant to urban planning, transportation engineering, regional science and economics.

Course Objectives and Format: This course covers the theory and advanced methods of choice modelling, with applications drawn from travel behaviour analysis in emerging urban settings with new modes such as shared mobility, ride hailing, integrated multimodal transit, and EVs. Assuming familiarity with the basic choice modelling methods such as the multinomial logit and maximum likelihood estimation, the course will quickly delve into advanced model formulations and estimation methods. Equal emphasis will be given to empirical model specification and behavioural interpretation issues, including causality, behavioural heterogeneity, and endogeneity. In addition to theories and modelling methods, importance is given to hands-on estimation, specification, and interpretation of choice models on real-life empirical datasets using open-source model estimation tools. Lectures in the course will include a mix of traditional, chalk-and-board sessions and PowerPoint presentations. The lectures will be accompanied with software laboratory sessions to provide hands-on experience with model formulations and estimation methods. Course participants need to bring their laptops for hands-on software laboratory sessions involving estimation and interpretation of choice models. No computers or laptops will be provided at the site.

Course Contents:

• Brief review of utility-based choice theories, multinomial logit, and maximum likelihood estimation


• Model specification, behavioural interpretation, causality, and endogeneity issues


• Data collection and survey design aspects for choice modelling


• Advanced choice models with cross-sectional & panel data, including psycho-social attitudinal and life-style variables


• Multivariate & stochastic variable models (models with multiple dependent variables, models with stochastic variables)


• Parameter identification in basic and advanced choice models


• Discrete-continuous choice models and multiple discrete-continuous choice models (MDCEV, etc.)


• Advanced estimation methods (simulation-based estimation, methods for approximating multivariate normal integrals, etc.)


• Modelling lab sessions with estimation, interpretation, and specification exercises

Prerequisites:

Course participants are expected to be comfortable with matrix algebra; differential and integral calculus; probability and statistics; estimation and interpretation of linear regression models and hypothesis testing; maximum likelihood estimation; and the basic multinomial logit model.

Who will benefit from the course:

• The course is targeted to those who are already familiar with the basic choice modelling methods such as the multinomial logit and maximum likelihood estimation.
• Post graduate students and research scholars working in fields of Transportation Engineering and Planning, Urban Planning, Geography, Economics, Management Sciences, Data Analytics and Machine Learning.
• Faculty members and research staff from academic institutes and R&D centres.
• Practicing professionals from consulting firms and transportation planning and transit agencies.

Course Dates:

3rd - 7th February 2020 (begins at 9am on 3rd February and ends at 5pm on 7th February)

Venue:

Centre for Continuing Education (CCE), IISc Bangalore

Registration

Registration is necessary. Please use the following link (and select this course) to register and pay fees online: Register here

Please upload a PDF of your resume/CV at the end of the online registration form. Last date for registration is 20th January 2020. Number of participants for the course is limited to 40. Registration fees are :

• Students from Indian academic institutions: Rs. 4,000/- (plus 18% GST). Students must bring college ID to the course
• Non-student participants from Indian academic institutions and R&D units: Rs. 6,000/- (plus 18% GST)
• Participants from Indian industry entities: Rs. 10,000/- (plus 18% GST).
• Students from academic institutions outside India: US $100 for low-income countries, US $150 for lower-middle-income countries, US $200 for upper-middle-income countries, US $400 for high-income countries. Fees include taxes.
  
• Industry and other participants from outside India: US $200 for low-income countries, US $250 for lower-middle-income countries, US $350 for upper-middle-income countries, US $600 for high-income countries). These fees include taxes. Country income classifications according to the World Bank 2019 definitions are at the link below:
  https://datahelpdesk.worldbank.org/knowledgebase/articles/906519-world-bank-country-and-lending-groups

The fee includes lunch and two tea/snack breaks on all three days. The course fee does not include accommodation.

Participants are advised to bring their own laptops for hands-on exercises involving model estimation and interpretation.

Accommodation for Outstation Participants: Accommodation is not guaranteed, but efforts will be made to reserve a limited number of rooms for outstation participants. Participants need to request in advance (no later than 20th January 2020) for accommodation at the registration website. Room rent may be paid later, directly to the accommodation facility.

Course Faculty:

• Prof. Chandra Bhat (University of Texas at Austin) Profile
• Prof. Ram Pendyala (Arizona State University) Profile
• Prof. Karthik Srinivasan (IIT Madras) Profile
• Prof. Abdul Rawoof Pinjari (IISc Bangalore) Profile
• Other invited experts from academia and industry.

Acknowledgements:

This short course is part of a SPARC project titled “Synergistic formulation of smart mobility solutions for India: “Mobility as a service (MaaS), electric vehicles (EVs), & integrated public transit” funded by the Ministry of Human Resource Development, Government of India.

Venue & Contact