Skip to main content

Our central research topic is the development and use of data mining techniques for a better decision making process. ADM has two main research lines:

  1. Explainable AI: explaining the decisions made by predictive models. Either global models that explain the model over the complete input space (e.g. through rule extraction) or instance-based explanations that explain individual predictions (e.g. through evidence counterfactual/EDC)
  2. Mining behavioral data: behavioral data is the collection of breadcrumbs that persons leave behind as they perform actions that are recorded digitally. Examples include visiting particular locations, visiting webpages, liking Facebook pages and purchasing products. We develop methods tailored to such data, which is sparse, ultra-high dimensional and typically class-imbalanced, and apply these in various applications (from digital advertising to fraud detection).

Application domains

  • Marketing: digital advertising, response modeling, churn prediction
  • Financial risk management: fraud detection and credit scoring