Resources

Textbooks

An Introduction to Statistical Learning — James, Witten, Hastie, Tibshirani (2021, Springer). The spine of the course. Chapters 3, 4, 6, 8, 10 and 12 map almost one-to-one onto our lectures. Free PDF + labs →

Data Science for Business — Provost & Fawcett (2013, O’Reilly). Read this one for why, not for how. Best source on framing a business problem as a data problem, and on what a model is actually worth to a firm.

Practical Time Series Forecasting with R — Shmueli & Polak (2024, Axelrod Schnall). Companion for the Module 1 forecasting block.

Additional readings and support material: Blackboard.

Materials on Blackboard

Two folders are open on Blackboard, one per module.

They are visible but empty for the moment — deliberately so, so that you can find them and check they are there before there is anything to download. Prof. Arbia’s material fills the Module 1 folder and Prof. Salvini’s the Module 2 folder, posted as each block runs rather than all at once.

Slides, code and datasets linked from the schedule are the same files you will find in those folders. Nothing is published in one place only.

Software

Installing R, RStudio and Python, the RStudio settings that save you a semester of grief, and the full list of course packages: Setup.

Going deeper

Optional. None of this is examinable — it is here for when a topic catches you.

Regularisation and high-dimensional data
Statistical Learning with Sparsity — Hastie, Tibshirani, Wainwright. Free PDF
Forecasting
Forecasting: Principles and Practice (3rd ed.) — Hyndman & Athanasopoulos. Free online — the reference for fable.
Model interpretability
Interpretable Machine Learning — Christoph Molnar. Free online — SHAP, LIME, partial dependence, done properly.
Data visualization
Fundamentals of Data Visualization — Claus Wilke. Free online
The Visual Display of Quantitative Information — Tufte. Buy it on paper.
Generative AI in the analytics workflow
Use it. Then check it. A model that writes glmnet code will also confidently write glmnet code that standardises the wrong matrix — the exam asks you to read code, not to have generated it.

Datasets

Datasets used in the labs land in materials/ and are linked from the schedule.