Quantitative probing: Validating causal models with quantitative domain knowledge

Abstract We propose quantitative probing as a model-agnostic framework for validating causal models in the presence of domain knowledge. The method is constructed analogy to train/test split correlation-based machine learning. It consistent with logic scientific discovery and enhances current validation strategies. effectiveness illustrated using Pearl’s sprinkler example, before thorough simulation-based investigation conducted. Limits technique are identified by studying exemplary failing scenarios, which furthermore used list topics future research improvements presented version probing. A guide practitioners included facilitate incorporation modelling applications. code integrating into analysis, well studies provided two separate open-source Python packages.