Advanced Quantitative Research Methods¶
A complete, ready-to-teach university course in statistical learning — ten Beamer decks, ten Jupyter notebooks, nine advanced modules, eight mock exams and the course datasets, sharing one notation and one semester rhythm.
11 decks 901 core slides (+199 optional) 109 exercises with solutions 11 labs (with solutions) 8 advanced modules 3 + 5 mock exams 23 datasets
Prepared by Prof. Dr. Christoph Weisser.
Based on ISLP
These materials follow An Introduction to Statistical Learning, with Applications in Python (James, Witten, Hastie, Tibshirani & Taylor, Springer 2023 — “ISLP”). The structure, topics, notation and labs follow the book; please cite it if you reuse them — see Citation & licence.
Start here¶
Prerequisites, workload, the skip rule, how to revise — and Colab on day one.
Semester plan, per-session runsheets, slide index, one make command.
LaTeX sources, generated figures, pinned environment — all editable.
The materials¶
The 11-session plan, the chapter map, the three split points — and how the course is graded.
Ten decks: 901 core slides, 199 more in optional appendices, every exercise with a worked solution.
Ten labs, one per deck, each closing with worked solutions — rendered here in full and runnable on Colab or locally.
Three full-length papers plus five 60-minute short exams — documented here, distributed on request.
Run a lab in Colab with zero setup; install locally from week two.
The 23 ISLP datasets bundled with the course, with sizes and where each is used.
What is pinned, why, and which chapter needs which extra package.
Nine optional self-study modules: RCTs, Shapley values, conformal prediction, GLMs & splines, SVMs, survival analysis, multiple testing, unsupervised learning, deep learning.
Six 3–5 hour challenges: a real decision on real data, ending in a memo.
What makes these materials different¶
- A whole course, not a pile of files.
Decks, labs and exams that share one notation, one dataset set and one semester rhythm — ready to teach as-is or adapt.
- Slides built for the room.
Every deck moves motivation → intuition → formal definition → worked example, with colour-coded callout boxes and 74 short + 35 extended exercises, each followed by a full solution. The hardest, optional material sits in a per-deck appendix, so the main thread fits the sessions it has.
- Numbers you can trust.
71 purpose-built figures are computed from the real course datasets (not sketched), 38 more are drawn natively in TikZ, and every mock-exam answer was verified programmatically.
- Ready to walk into a room.
Timed runsheets, a generated slide index with page numbers, a ranked cut list for when you are behind, and
make checkto catch a slide that overruns its frame.- Reproducible by design.
LaTeX sources for every deck and exam, a pinned Python environment, and datasets that resolve automatically — locally and on a fresh Colab runtime.
At a glance¶
Lecture decks |
8 taught ISLP chapters + two precourse decks · 901 slides, plus 199 in optional appendices |
Exercises |
74 short (~5 min) + 35 extended (~15 min), all with worked solutions |
Lab notebooks |
10 labs, one per deck (2 precourse + 8 chapters), each closing with worked Python solutions; several also verify the deck’s by-hand arithmetic in code |
Mock exams |
3 full-length papers, each in 3 formats, + 5 sixty-minute short exams (not distributed publicly) |
Datasets |
23 CSVs from statlearning.com |
Advanced modules |
9 optional self-study modules (RCTs, Shapley values, conformal prediction, GLMs & splines, SVMs, survival analysis, multiple testing, unsupervised learning, deep learning) · 688 core slides (+89 in appendices) + 9 notebooks |
Short projects |
6 challenges on real data, 3–5 h each, each with a fixed held-out set, a baseline to beat and a memo as the deliverable |
Semester shape |
11 × 180 min: a taught precourse session + 11 chapter sessions · 6 ECTS |
Assessment |
One written exam at the end of the semester (120 min, 100% of the grade); the eight practice papers do not count |
Sources |
LaTeX (Beamer) · Jupyter · Python 3.9+ · one |