Scientific method development with accessible publications and software.
Explainable AI
Thrun et al. · PLOS Medicine · 2026
Structure discovery, explicit classification logic and assessment of individual predictions are combined for lymphoma classification.
Retrospective evaluation uses 19,493 blood samples and external data. The publication, archived code and PLAIT demonstrator can be accessed separately.
Temporal Data Mining
Stier, Gehlert & Thrun · Multiresolution Forecasting for Industrial Applications · Processes · 2021
The paper investigates wavelet-based multiresolution forecasting for industrial time series. It combines features at different time scales with systematic comparison of forecasting methods.
The study compares forecasting methods on four datasets and multiple horizons using rolling forecasting origins. Errors are assessed with MASE and SMAPE. The method is available in R and Python, including through the mrf R package.
Examine data structure and quality
Classify and explain the result
Assess reliability for the individual case
Review the result with domain expertise
Stier, Q., Hoffmann, J. & Thrun, M. C. (2026). Classifying with the Fine Structure of Distributions: Leveraging Distributional Information for Robust and Plausible Naïve Bayes. Machine Learning and Knowledge Extraction (MAKE), 8(1), 13. MDPI.
A methodologically grounded starting point for classification: PDEnaiveBayes uses the fine structure of data distributions to assign classes. As a baseline, it helps assess the additional value that more complex models offer for a particular task. Pareto density estimation and a plausibility correction provide the methodological basis. The publication and R package on CRAN support reproducible comparisons on your own data.
Thrun, M. C. (2022). Exploiting Distance-Based Structures in Data Using an Explainable AI for Stock Picking. Information, 13(2), 51. MDPI.
The paper connects distance-based structures in company fundamentals with explainable decision rules for stock selection. The application illustrates how knowledge discovery can provide a basis for decisions that domain experts can assess.
Thrun & Ultsch · Artificial Intelligence · 2021
Swarm intelligence and self-organisation for clustering high-dimensional data. Connects methodological research with a publicly available implementation.
Thrun, Ultsch & Breuer · Machine Learning and Knowledge Extraction · 2021
The hydrochemical application connects clustering with interpretable states and explanations. An example of temporal knowledge discovery that goes beyond predicting a single numerical value.
Thrun, M. C. & Märte, J. (2026). Memshare: Memory Sharing for Multicore Computation in R with an Application to Feature Selection by Mutual Information using PDE. The R Journal, 17(4), 306–322.
Shared memory for parallel computation in R. Parallel R processes access the same data without each making a full copy.
From data exploration and clustering to classification, forecasting and efficient processing. Each package provides documentation, authorship and licensing terms.
Explore distributions and data
Thrun, M. C., Gehlert, T. & Ultsch, A. (2020).
Analyzing the fine structure of distributions ↗
PLOS ONE, 15(10), e0238835.
Discover structures through self-organisation
Thrun, M. C. & Ultsch, A. (2021).
Swarm intelligence for self-organized clustering ↗
Artificial Intelligence, 290, 103237.
Projection-based clustering
Thrun, M. C. & Ultsch, A. (2021).
Journal of Classification, 38(2), 280–312.
Classification using distributional information
Stier, Q., Hoffmann, J. & Thrun, M. C. (2026).
Machine Learning and Knowledge Extraction (MAKE), 8(1), 13.
Multiresolution forecasting
Stier, Q., Gehlert, T. & Thrun, M. C. (2021).
Multiresolution Forecasting for Industrial Applications ↗
Processes, 9(10), 1697.
Time-series analysis
Thrun, M. C., Märte, J., Böhme, P. & Gehlert, T. (2019).
Applying Two Theorems of Machine Learning to the Forecasting of Biweekly Arrivals at a Call Center ↗
European Conference on Data Analysis (ECDA), conference presentation.
Shared memory for parallel analysis
Thrun, M. C. & Märte, J. (2026).
The R Journal, 17(4), 306–322.
Clustering and estimation of the number of clusters
In a free, non-binding 20-minute introductory discussion, we clarify your objective, starting point and a useful next step.
