Winter School with the Expert: Causal Inference with Machine Learning

What:
This is an advanced inclusive training in causal inference. The course will comprise personalised training on fundamental methods in mathematics and statistics, introduction of causal inference and to machine learning, and three days of more advanced specialist content, delivered by Professor Jannis Kuck. The diagnostic tests, administered in advance of the course, will identify participants’ pre-existing knowledge of relevant quantitative skills. Participants will receive personalised timetables for the course, including online teaching material for basic prerequisite knowledge, pointing them to training in quantitative methods where needed, as identified by analysis of the test results.

There will be pre-requisite courses in December 2025 on: OLS ‘refresher’, Introduction to R and Visualization, Introduction to Causal Inference, Use of Panel Data for Causal Inference and data handling. These will be supported by Amruta Bagwe, Daniel Cernin, Armine Ghazaryan and Valentina Di Iasio.

By the end of the Winter School, you will:
• Have gained a clear understanding of the relationship between causal inference and machine learning, and how these approaches differ from purely regression analysis
• Understand the main approaches to causal inference via modern machine learning techniques for both linear and nonlinear models
• Be able to apply and critically assess common strategies in applied/empirical causal analysis
• Have acquired hands-on experience in implementing machine learning methods for estimating causal effects using R
• Be equipped with the background knowledge to engage with more advanced causal machine learning methods beyond the scope of this course.

When:
The course starts with a diagnostic test early in December 2025; hybrid ‘bridge’ courses to cover pre-requisites with short videos, lectures and surgery sessions (first two weeks in December). This is followed by 3-days in person on Advanced Causal Inference:
7 January 2026 11.00-17.00
8 January 2026 11.00-17.00
9 January 2026 9.00-15.30

Where:
Online and University of Southampton, Highfield Campus, Building 100, Room 6009

The expert:
Professor Jannis Kuck, Professor of Economics and Data Science at the University of Düsseldorf, specialises in high-dimensional statistics, econometrics, causal inference, machine learning and graphical models to provide an inclusive course to Social Sciences PhD students. Jannis’ expertise both in high dimensional statistics and in the frontier topic of causal ML, combined with the in-house expertise of Dr Chaowen Zheng guarantees the delivery of state-of-the-art training in this area.
https://www.dice.hhu.de/en/dice/people/professors-1/kueck and https://www.southampton.ac.uk/people/656hss/doctor-chaowen-zheng

Who is the training for:
The training is suitable for PhD students in Quantitative Social Sciences. Pre-requisites for the 3-day course in January, which will be offered as part of the package are:
1. Basic understanding of probability theory (expectations etc), statistical theory (hypothesis testing), and regression analysis (OLS) refresher
2. Basic understanding of causal analysis (randomized experiments, confounding factors etc)
3. Basic experience with data analysis using software (Stata or R)—We will use R, but experience with Stata might also help
4. Not required, but an advantage: Basic understanding of Machine Learning methods, in particular shrinkage methods (e.g., Lasso, Ridge) and tree-based methods (regression trees, random forest)

Details of the course (to be finalised):
Day 1
Topic 1 Morning: Recap of Regression analysis/ Causal inference/ Machine Learning; Topic 2 Afternoon: Causal inference in high dimensional linear model (mainly Lasso) and its application using R.

Day 2
Topic 3 Morning: Causal Instrumental analysis in high dimensional linear model (mainly double machine learning) and its application using R; Topic 4 Afternoon: Causal inference via modern nonlinear model (mainly random forest) and its application using R.
Day 3 Topic 5 Morning: Causal DiD analysis and its application using R; Afternoon: WORKSHOP: Mapping the actual state of causal machine learning with relevant research questions, plus discussion on any research related issues.

How to join: sign up here
This training is now full.