Course – Discrete Choice Analysis: micro-econometrics and machine learning approaches

July 6, 2022 10.00 - 16.00 h

General aim
Discrete choice analysis (DCA) has become one of the most important frameworks for transportation modelling. Using DCA, the researcher or analyst is able to estimate the influence of all sorts of factors on travel choice behavior, and to predict mobility patterns and market shares for transport-related services; all this, in a quantitative, statistically rigorous way with deep roots in economics and the behavioral sciences. As such, DCA is indispensable for the underpinning of many transport policies and plans.

In this course, we will cover two different perspectives on DCA: the conventional, econometrics-based perspective (also called Discrete choice theory); and a more novel perspective which is gaining ground rapidly, based on recent advances in Machine learning. This combination makes this course unique, compared to other choice modelling courses taught in the Transport community.

This course will contain a mix of theory, implementation guidelines, and hands-on exercises to be completed during the course and under supervision of the lecturer.

Participants of this course will be awarded with 2 ECTS when attending the full course (one for each part).

This course covers:
Day 1 – Basics of Discrete choice theory (Chorus)

  • Specification and estimating a discrete choice model
  • The Logit-model based on Random Utility theory
  • Economic appraisal with discrete choice models
  • Exercise 1 (Apollo)

Day 2 – Advances in Discrete choice theory (Chorus)

  • The (panel) Mixed Logit-model (error components and random parameters)
  • Random regret minimization-based Discrete choice theory
  • Exercises 2 and 3 (Apollo)

Day 3 – Introduction to Machine learning for choice behaviour analysis (Van Cranenburgh)

  • Introduction into Machine learning for choice behaviour analysis
  • Data and training
  • Artificial Neural networks, SVM, Random Forests
  • Exercise 4 (Jupyter notebook)

 Day 4 – Advances in Machine learning for choice behaviour analysis (Van Cranenburgh)

  • Hybrid models
  • Explainable AI techniques
  • Exercise 5 (Jupyter notebook)

Course leaders: Prof. Caspar Chorus and Dr. Sander van Cranenburgh

Dates: 5, 6, 7, 8 July 2022

Time: 10.00 – 16.00 h

Location: TU Delft (on-site)

Faculty: TPM, Jaffalaan 5

Room: D2 (ground floor)

ECTS: 2 (attendance, including assignments)

Registration: please CLICK HERE

Participation: free for TRAIL/Beta/OML/ERIM members and PhD students of participating TRAIL faculties. Others please contact the TRAIL Office

© 2020 Research School TRAIL