ChatGPT, Claude, and other generative AI applications are quick to open, but they only deliver truly excellent results once you understand how they work. As an introduction, we explain what language models are and why they sometimes give convincing-sounding but incorrect answers. Next, we focus on effective prompting: clearly describing tasks, providing examples, and improving results step by step. In the exercises, participants draft emails, summarize long documents, extract information from PDFs, and analyze tables. They also learn to critically evaluate results and recognize where the AI reaches its limits. By the end of the day, everyone has a collection of best practices they can immediately apply in their daily work.

Machine learning powers many AI applications, but how does a model actually learn from data? This workshop explains the core concepts clearly and without unnecessary formulas: regression and classification, decision trees, and random forests. A key focus is on properly evaluating models. This includes questions such as how to measure predictive accuracy, what overfitting is, and how resampling techniques like cross-validation protect against it. The workshop also addresses typical pitfalls that cause ML projects to fail in practice. All methods are tested directly in R or Python using realistic datasets, enabling participants to train, compare, and evaluate simple models themselves afterward.

High-performance machine learning models are often a black box: they deliver good predictions, but no explanations. In this workshop, participants will learn methods for understanding the behavior of such models. Global methods like Feature Importance and Partial Dependence Plots reveal which factors a model uses overall. Local methods like Shapley Values or Counterfactual Explanations, on the other hand, explain individual predictions. We will also show where these methods reach their limits and how to avoid misinterpretations. All methods will be applied practically in R or Python, and the workshop will conclude with a focus on communicating the results clearly to business departments, management, or clients.

This intensive week is designed for anyone who wants to confidently implement machine learning (ML) projects beyond the basics. It covers the entire project lifecycle, from data exploration and preparation to feature engineering, model selection, and hyperparameter tuning. A key focus is on the most common errors encountered in practice: data leakage, imbalanced classes, missing values, and evaluations that produce overly optimistic results. We demonstrate how to identify such problems and avoid them with clean pipelines. Participants will work with various datasets to ensure each topic is illustrated with a relevant example. Afterward, participants will be able to approach their own ML projects in a structured manner and avoid typical pitfalls.

All our formats can also be tailored to your company. Instead of general case studies, we work with examples from your industry, and if desired, with your own data. Content can be combined, expanded upon, or supplemented with topics not covered in our standard formats. A specific use case from your company can also be the focus, resulting in an initial prototype. In a preliminary consultation, we clarify objectives, prior knowledge, and the appropriate scope. This leads to a program perfectly suited to your team.