Event

Most scientific questions — particularly those concerning the effects of policies, interventions, or exposures (henceforth referred to as ‘treatments') — are inherently causal, even when not explicitly framed as such. Causal inference provides a formal framework for defining these questions precisely and identifying the assumptions required to express them in terms of observable data. Once the causal question has been formulated, attention turns to estimation and statistical inference. Estimating causal effects typically requires adjusting for confounding. Traditional approaches for confounding adjustment rely on parametric statistical models, but these can perform poorly in the presence of high-dimensional confounding and are vulnerable to model misspecification bias. Machine learning methods are therefore increasingly used to support causal effect estimation. However, the use of machine learning in causal inference presents unique challenges and their naïve application can introduce bias.

This workshop provides an introduction to principled methods for incorporating machine learning in causal effect estimation, for both point treatments (treatments administered at a single time point) and time-varying treatments.

Day 1 will introduce causal reasoning, machine learning for prediction and essential causal inference methods, before highlighting how bias can arise when standard prediction-oriented methods are applied naively.

Day 2 will present principled machine learning approaches for estimating the causal effects of point treatments, specifically targeted learning and related debiased machine learning methods.

Day 3 will present the extension of these methods to time-varying treatments, for those interested in this more advanced setting.

The workshop can be attended as a 3-day workshop or as a 2-day workshop (Days 1 and 2 only), with all days consisting of a combination of lectures and hands-on R sessions.

Intended audience: The target audience is statisticians and researchers with some statistical background, including some knowledge of regression methods. For the computer practical, students must also have a sound working familiarity with R and have it installed in their laptop, which they must bring to the workshop.

Dates:

Wednesday 10 - Friday 12 February

Location:

The University of Melbourne

Venue location to be advised.

While in-person attendance is highly recommended, this course is also accessible remotely via Zoom.

Registration:

Registration will open soon and will close one week prior (3 February 2027) unless sold out.

Early-bird rates (a 10% discount on the below listed prices) are available until 30 September 2026.

Registration pricing:

Cancellation policy:

If you are no longer able to attend, please notify us at your earliest convenience.

Ticket changes and refunds may be requested up to Friday, 29 January 2027 by writing to .