Interpretable AI for travel behaviour: Embedding theory in machine learning

About the Project

Artificial intelligence (AI) and machine learning are increasingly used to forecast travel behaviour, offering impressive predictive capacity. Yet many such models lack grounding in behavioural theory and operate as “black boxes” that obscure how or why decisions are made. Classical behavioural models, such as discrete choice and hybrid choice frameworks, are built on solid behavioural foundations and explicitly represent concepts such as preferences, attitudes, and heterogeneity. However, they may struggle with complex, non-linear data.

This PhD seeks to combine the best of both: hybrid models that retain behavioural integrity while harnessing the flexibility of AI. The project will begin with a critical review of hybrid approaches across transport and related disciplines, focusing on how behavioural assumptions (e.g. utility structures, latent variables, preference variation) can be embedded into predictive architectures. You will then design and test hybrid models using empirical travel behaviour data (e.g. mode choice, route choice, or demand forecasting).

These models will be assessed not only for predictive performance, but also for how well they maintain behavioural realism, reveal underlying mechanisms, and capture diversity across people and contexts. The methodological focus and dataset can be tailored to your interests, ensuring flexibility within the overall framework.

The expected contribution is twofold. Methodologically, the project advances the design of AI systems that are behaviourally grounded, ensuring models both predict well and explain faithfully. Practically, it develops tools for planners and policymakers that support real-world decisions - from building more equitable and inclusive transport systems to enabling sustainable mobility transitions. By keeping models both accurate and trustworthy, this work helps ensure that technical innovation contributes directly to pressing societal challenges.

Academic qualifications

A first degree (at least a 2.2) ideally in Transport studies / transport planning, Economics (especially econometrics or applied microeconomics) Statistics, mathematics, or data science, Computer science (with focus on AI, machine learning, or modelling) with a good fundamental knowledge of the solid grounding in statistical and/or econometric modelling.

English language requirement

IELTS score must be at least 6.5 (with not less than 6.0 in each of the four components). Other, equivalent qualifications will be accepted. Full details of the University’s policy are available online.

Essential attributes:

• Hands-on experience with machine learning methods (e.g. classification, prediction, ensemble models)
• Awareness of behavioural theories of decision-making (such as utility, bounded rationality, or preference heterogeneity)
• Ability to implement and test statistical and machine learning models using a programming language such as Python, R, or MATLAB
• Capacity to think critically about the strengths and limitations of different modelling approaches
• Clear communication skills, particularly in explaining technical findings to non-technical audiences.
• Self-direction and persistence in developing and testing new methodological approaches

Desirable attributes:

• Prior exposure to discrete choice modelling, econometrics, or similar frameworks
• Knowledge of transport behaviour datasets or related applied behavioural data
• Interest in policy-relevant issues such as model interpretability, fairness, or sustainable transport applications

APPLICATION CHECKLIST

• Completed application form
• CV
• 2 academic references, using the Postgraduate Educational Reference Form (download)
• Research project outline of 2 pages (list of references excluded). The outline may provide details about:
  1. Background and motivation of the project. The motivation, explaining the importance of the project, should be supported also by relevant literature. You can also discuss the applications you expect for the project results.
  2. Research questions or objectives.
  3. Methodology: types of data to be used, approach to data collection, and data analysis methods.
  4. List of references.
  The outline must be created solely by the applicant. Supervisors can only offer general discussions about the project idea without providing any additional support.
• Statement no longer than 1 page describing your motivations and fit with the project.
• Evidence of proficiency in English (if appropriate)

To be considered, the application must use

• the advertised title as project title

For informal enquiries about this PhD project, please contact Dr Achille Fonzone - a.fonzone@napier.ac.uk

References

Arkoudi, I., Krueger, R., Azevedo, C. L., & Pereira, F. C. (2023). Combining discrete choice models and neural networks through embeddings: Formulation, interpretability and performance. Transportation research part B: methodological, 175, 102783.
Vij, A., & Hess, S. (2025). Posterior inference of attitude-behaviour relationships using latent class choice models. arXiv preprint arXiv:2509.08373.
Wang, S., Mo, B., Zheng, Y., Hess, S., & Zhao, J. (2021). Comparing hundreds of machine learning classifiers and discrete choice models in predicting travel behavior: an empirical benchmark. arXiv preprint arXiv:2102.01130.