Call for Contributions: Hackathon on Interpretable Machine Learning for Longitudinal and Clustered Data

Researchers are invited to contribute to a multi-analyst study and in-person hackathon on Interpretable Machine Learning for Longitudinal and Clustered Data, hosted by the interdisciplinary research center Agile Prevention and Intervention Research (Agile PAIR) at TU Dortmund University. The initiative addresses a growing challenge in the social and behavioral sciences: modern datasets often contain complex dependencies, such as repeated measurements over time or hierarchical structures like students nested within schools. At the same time, researchers increasingly rely on machine learning methods that provide strong predictive performance but are often difficult to interpret.

The event brings together researchers interested in developing and comparing interpretable machine learning approaches suitable for such dependent data structures. The goal is to evaluate different methods through a coordinated multi-analyst study and to derive practical guidelines for applying supervised machine learning in the presence of stochastic dependencies.

Researchers are free to apply a wide range of interpretable or explainable approaches, including decision trees, regularized regression models, functional or additive models, interpretable ensemble methods, neural networks with explainability techniques, and surrogate modelling approaches.

Contact: agile-pair@tu-dortmund.de