| Course information | |
|---|---|
| Period | Block 2 |
| Timeline | February |
| Number of ECTS | 5 ECTS |
| Coordinators | Dr Beatriz Calzada Olvera |
| Lecturers | Dr Beatriz Calzada Olvera and Dr Alberto Gianoli |
| Methodology | Lectures, workshops |
Course description
As cities and societies become increasingly digitalised, policy analysis relies on large and complex datasets (e.g., administrative records, satellite imagery, sensors, and digital platforms). Machine-learning (ML) methods offer powerful tools for prediction and pattern detection, but they also raise challenges related to interpretation, validation, and their distinction from causal inference.
This course provides students with a solid foundation in statistical learning for policy-relevant analysis, explicitly bridging econometric inference and machine-learning approaches. Students learn when ML methods are appropriate, how they differ from causal approaches, and how they can be responsibly applied in policy and urban research contexts.
The course builds on a shared quantitative core – multiple regression, limited dependent variable models, panel data, and instrumental variables – before introducing regularisation methods, tree-based models, and validation techniques. Emphasis is placed on critical assessment, academic application, and policy relevance, rather than purely technical optimisation.
Learning objectives
After successful completion of the course, students will be able to:
Common learning outcomes:
- Apply and design sound multivariate econometric models.
- Distinguish clearly between statistical inference, predictive accuracy, and causal inference.
- Communicate quantitative results clearly to non-technical, policy-oriented audiences.
Track-specific learning outcomes:
- Understand and apply regularised regression and tree-based machine-learning models, and evaluate predictive performance using appropriate validation techniques.
- Critically assess the suitability and limitations of machine-learning methods in policy contexts.
Teaching and learning methods
- Lectures (conceptual foundations, interpretation, and methodological trade-offs)
- Workshops (hands-on empirical exercises in R)
The course follows a lecture–workshop format, with workshops closely aligned to lecture content.
Practical exercises
- Replication and extension of empirical examples using real-world policy datasets
- Guided coding exercises in R (regression, regularisation, tree-based models)
- Interpretation exercises focusing on model explainability and policy implications
Exercises are formative and directly prepare students for the final assessment.