@inproceedings{60680,
  abstract     = {{Classical machine learning techniques often struggle with overfitting and unreliable predictions when exposed to novel conditions. Introducing causality into the modelling process offers a promising way to mitigate these challenges by enhancing predictive robustness. However, constructing an initial causal graph manually using domain knowledge is time-consuming, particularly in complex time series with numerous variables. To address this, causal discovery algorithms can provide a preliminary causal structure that domain experts can refine. This study investigates causal feature selection with domain knowledge using a data center system as an example. We use simulated time-series data to compare 
different causal feature selection with traditional machine-learning feature selection methods. Our results show that predictions based on causal features are more robust compared to those derived from traditional methods. These findings underscore the potential of combining causal discovery algorithms with human expertise to improve machine learning applications.}},
  author       = {{Zapata Gonzalez, David Ricardo and Meyer, Marcel and Müller, Oliver}},
  keywords     = {{Causal Machine Learning, Causality in Time Series, Causal Discovery, Human-Machine  Collaboration}},
  location     = {{Amman, Jordan}},
  title        = {{{Bridging the gap between data-driven and theory-driven modelling – leveraging causal machine learning for integrative modelling of dynamical systems}}},
  year         = {{2025}},
}

@article{34114,
  abstract     = {{Qualitative comparative analysis (QCA) enables researchers in international management to better understand how the impact of a single explanatory factor depends on the context of other factors. But the analytical toolbox of QCA does not include a parameter for the explanatory power of a single explanatory factor or “condition”. In this paper, we therefore reinterpret the Banzhaf power index, originally developed in cooperative game theory, to establish a goodness-of-fit parameter in QCA. The relative Banzhaf index we suggest measures the explanatory power of one condition averaged across all sufficient combinations of conditions. The paper argues that the index is especially informative in three situations that are all salient in international management and call for a context-sensitive analysis of single conditions, namely substantial limited diversity in the data, the emergence of strong INUS conditions in the analysis, and theorizing with contingency factors. The paper derives the properties of the relative Banzhaf index in QCA, demonstrates how the index can be computed easily from a rudimentary truth table, and explores its insights by revisiting selected papers in international management that apply fuzzy-set QCA. It finally suggests a three-step procedure for utilizing the relative Banzhaf index when the causal structure involves both contingency effects and configurational causation.
}},
  author       = {{Haake, Claus-Jochen and Schneider, Martin}},
  journal      = {{Journal of International Management}},
  keywords     = {{Qualitative comparative analysis, Banzhaf power index, causality, explanatory power}},
  number       = {{2}},
  publisher    = {{Elsevier}},
  title        = {{{Playing games with QCA: Measuring the explanatory power of single conditions with the Banzhaf index}}},
  volume       = {{30}},
  year         = {{2024}},
}