Given data, can you establish causation — not just correlation?

The Constructor University Datathon 2026 invites participants to tackle high-level challenges in causal reasoning and probabilistic modeling. Sponsored by Insytes GmbH, this three-day event challenges teams to apply theory—from directed acyclic graphs to correlation analysis—to practical industry problems. Participants will benefit from the direct support of Insytes GmbH and academic guidance from our specialized faculty.

📋 Table of Contents

Overview

Theme & Topic

This year's Datathon is centred on one of the most fundamental questions in data science: not just what is happening in the data, but why. The theme — Causation, Correlation & Causal Inference — challenges participants to move beyond surface-level patterns and build a rigorous understanding of cause and effect.

At the heart of the challenge is causal inference — a framework for reasoning about the effect of interventions and counterfactuals in complex systems. Participants will work with Directed Acyclic Graphs (DAGs) to map out causal relationships between variables, identify confounders, and determine what can and cannot be inferred from observational data alone.

Probabilistic reasoning ties it all together — giving teams the tools to quantify uncertainty, model dependencies, and draw conclusions that are not just statistically sound, but causally meaningful. Whether you are building models, interpreting results, or designing interventions, thinking probabilistically is key to getting the right answer for the right reasons.

Day 1 — April 17: Opening, Talks & Challenge Launch

Day 1 focuses on orientation, context-setting, and launching the challenge. The challenge introduction and technical briefing run back to back so teams have a complete picture before the keynote, academic talk, and panel discussion. Teams begin working in the afternoon.

Day 1 at a glance:

14:00–14:30	Registration & participant check-in
14:30–14:45	Opening remarks & event introduction
14:45–15:15	Keynote talk — Automated Theorem Proving Using Causal ML
15:15–15:35	Academic talk — Introduction to Causal AI
15:35–16:05	Panel discussion — From Proof to Product
16:05–16:35	Coffee break & networking
16:35–16:55	Introduction of the challenge problem
16:55–17:15	Technical briefing for teams
17:15–19:00	Team Work Session 1 — problem exploration

Day 2 — April 18: Team Work & Support

Day 2 is primarily a free working day for teams, focused on independent development of solutions. Teams formalize causal hypotheses using DAGs, build causal and probabilistic models, and iterate on testing and validation. Optional in-person support is available for teams who need help.

Day 2 takes place in the IRC Seminar Room, with lunch self-arranged.

Day 2 at a glance:

09:00–12:00	Team Work Session 2 — model development
11:00–13:00	Optional in-person Q&A and support session (IRC Seminar Room 3)
12:00–13:00	Lunch break (self-arranged)
13:00–17:00	Team Work Session 3 — final development, visualizations, and draft report/slides

Day 3 — April 19: Finalisation, Presentations & Awards

Day 3 is for finalization, submission, judging, and closing. Teams finalize code, report, and slides in the morning, then submit all deliverables by noon. Presentations and evaluation follow in the afternoon, ending with awards and closing networking.

Submission deliverables are: code repository, final report, and presentation slides.

Day 3 at a glance:

09:00–12:00	Team Work Session 4 — finalization
12:00	Submission deadline
12:00–13:00	Lunch break (self-arranged)
12:00–15:30	Judging setup (East Wing, on-site and remote evaluation)
15:30–17:00	Team presentations (6 minutes + Q&A per team)
17:30–18:00	Judging & deliberation
18:00–18:20	Awards ceremony & closing remarks
18:20–19:00	Closing dinner & networking

Judges & Speakers

Prizes & Awards

Sponsors & Partners

The Datathon 2026 is organised with the support of Constructor University as the host institution and Insytes GmbH as the industry partner. Insytes GmbH contributes the challenge problem, dataset, mentorship, and judging — bringing real-world industry expertise directly into the competition.