Explainable Artificial Intelligence for Business Analytics


6 CP


8th May 2024
Location: Builidng D3 2, Room: HG 0.08






“Explainable Artificial Intelligence for Business Analytics" is class aimed at equipping students with the knowledge and skills necessary to comprehend, interpret, and convey AI-based decisions in business analytics settings. The course begins by introducing fundamental data-driven analytics topics, such as explorative data analysis, clustering, classification, and regression analysis. However, its primary focus is on the explainability of AI models. Students will learn about intrinsically interpretable machine learning approaches and black-box models, supplemented with a range of techniques for global and local post-hoc explanations. Emphasizing the ethical implications of AI, the course also underscores the need for responsible AI practices. By grasping the significance of explainability in AI and mastering the methods to achieve it, students will be well-prepared to ensure that their AI-driven analytics solutions are transparent, reliable, and trustworthy in real-world business situations, while promoting ethical and accountable decision-making processes.


Dr. Nijat Mehdiyev
E-Mail: nijatsamiloglu.mehdiyev[at]uni-saarland.de or nijat.mehdiyev[at]dfki.de
The course is offered by the Institute for Information Systems (Iwi) at Saarland University.


  1. Explorative Data Analysis with tidyverse (dplyr, tidyr, ggplot2)
  2. Clustering with tidymodels (tidyclust) and h2o
  3. Classification with tidymodels (rsample, recipes, parsnip, tune, yardstick) and h2o
  4. Regression with tidymodels (rsample, recipes, parsnip, tune, yardstick) and h2o
  5. Intrinsically Interpretable Machine Learning Approaches
  6. Black-Box Machine Learning Models
  7. Global Post-Hoc Explanations with vip, DALEX, kernelSHAP
    1. Permutation Feature Importance
    2. Partial Dependence Plots (PDP)
    3. SHAP Summary Plots/SHAP Dependence Plots
  8. Local Post-Hoc Explanations with vip, DALEX
    1. Individual Conditional Expectation (ICE) Plots
    2. SHapley Additive Explanations (SHAP)
    3. Local Surrogate Models
    4. Counterfactual Explanations
  9. Evaluation of XAI Methods

Date and time

The class will start at the beginning of May 2024 and run for 6 weeks


Assignments, group project and presentation


If we have aroused your interest, please register by sending your contact information and a short message to nijat.mehdiyev(at)dfki.de.