Explainable AI (XAI) – training for companies
Powerful ML models deliver good predictions, but rarely a reason. At the latest when specialist departments, management, customers or auditors ask why the model decided as it did, you need sound answers. In this two-day course, participants learn to make the behaviour of ML models understandable and to interpret the results correctly.
Who the course is for
-
Data scientists and ML developers who need to explain their own models
-
Analysts who present model results to specialist departments
-
Teams in risk management, quality assurance or compliance with a technical background
Content
-
Why explain models: black-box models, trust and transparency
-
Global methods: feature importance and partial dependence plots show which variables a model relies on overall
-
Local methods: Shapley values and counterfactual explanations justify individual predictions
-
Limits and pitfalls: when explanations mislead, for example with correlated features, and how to avoid misinterpretations
-
Communicating results: presenting explanations clearly for specialist departments, management or customers
All methods are applied hands-on in R or Python.
What participants can do afterwards
-
choose suitable explanation methods for their question
-
compute global and local explanations themselves and interpret them correctly
-
recognise and avoid typical misinterpretations
-
present results so that non-experts understand them too
Why us
Fiona K. Meinzer is doing her PhD at the Department of Statistics at LMU Munich on explainable AI, in particular on feature importance methods. She is a co-author of the guide “A Guide to Feature Importance Methods for Scientific Inference” (XAI 2024), of the paper “CASHomon Sets” (ECML 2026), which finds, across several model classes and their hyperparameters, the models that perform almost as well as the best one, and of the R package xplainfi (The R Journal, 2026), published partly under her birth name Ewald. In this course, you learn the methods from someone who develops them further herself, including the points where they reach their limits.
Framework
2 days
Duration: 2 days, on-site or online (on request in 90-minute sessions) Language: German or English Included: slides, exercises with solutions in R or Python and a confirmation of participation
Prerequisites: ML fundamentals (e.g. from our 2-day course) and confident use of R or Python. Laptop with R or Python.
€7,800
Equivalent to €650 per person per day for 6 participants
Also available as a custom workshop
1–5 days
On request, we explain your own models and data or combine XAI with other topics. In a preliminary call, we clarify goals, prior knowledge and scope. The custom format is offered separately.
Prerequisites: depending on the topic; we clarify this in the preliminary call.
from €3,300 per day plus workshop design
From €550 per person per day for 6 participants
All prices are net, excluding VAT and travel expenses. Prices are valid for in-house training courses with up to 12 participants; larger groups are available upon request. We recommend keeping groups small to allow sufficient time for questions and individual exercises. All formats are also available online, and can be divided into 90-minute sessions upon request.