Opening the Black Box

Interpreting Machine Learning Models with SHAP Values in R

A 90-minute tutorial at uRos2026 (Use of R in Official Statistics), Paris, 18 November 2026.

Many machine learning models are accurate but opaque. This tutorial opens the black box with SHAP values: what they are, how they’re computed, and how to read them; with far more practice than theory.

Where to go

Getting set up

Two set-ups are possible: on your own machine (clone the repo, restore the environment) or on the RStudio Server. A local install is best done before the session.

Open the setup guide →

Slides

The deck used during the live session: light on text, built around interactive widgets, an app and coding exercises.

Open the slides →

Course notes

The full reference document (theory, code, context…), not narrated live. Feel free to read it before or after the session, at your own pace.

Open the notes →

Further reading

• DALEX documentation: https://modeloriented.github.io/DALEX/: the explain()-based tools (feature importance, PDP, break-down and SHAP), used throughout the tutorial
• DALEXtra documentation: https://modeloriented.github.io/DALEXtra/: the extension of DALEX that provides explain_tidymodels(), the explainer used for tidymodels workflows
• shapviz documentation: https://modeloriented.github.io/shapviz/: the waterfall, beeswarm and dependence plots of the TreeSHAP part, with support for several other modelling packages beyond LightGBM
• kernelshap: https://CRAN.R-project.org/package=kernelshap: KernelSHAP for any model, including tidymodels workflows
• fairmodels: https://fairmodels.drwhy.ai/: fairness checks between protected groups, from the same DALEX explainer
• Molnar, Interpretable Machine Learning (Molnar 2025): the reference book, with chapters on SHAP, PDP and permutation importance
• Biecek and Burzykowski, Explanatory Model Analysis (Biecek and Burzykowski 2021): the book behind DALEX (break-down, SHAP, profiles)
• Lundberg et al. (Lundberg et al. 2020): TreeSHAP and global understanding of tree models
• Aas et al. (Aas et al. 2021): SHAP with dependent features

Aas, Kjersti, Martin Jullum, and Anders Løland. 2021. “Explaining Individual Predictions When Features Are Dependent: More Accurate Approximations to Shapley Values.” Artificial Intelligence 298: 103502. https://doi.org/10.1016/j.artint.2021.103502.

Biecek, Przemyslaw, and Tomasz Burzykowski. 2021. Explanatory Model Analysis. Chapman; Hall/CRC. https://ema.drwhy.ai/.

Lundberg, Scott M., Gabriel Erion, Hugh Chen, et al. 2020. “From Local Explanations to Global Understanding with Explainable AI for Trees.” Nature Machine Intelligence 2 (1): 56–67. https://doi.org/10.1038/s42256-019-0138-9.

Molnar, Christoph. 2025. Interpretable Machine Learning: A Guide for Making Black Box Models Explainable. 3rd ed. https://christophm.github.io/interpretable-ml-book.

See you in Paris! 🚀