Optimal Experiments for Partial Causal Effect Identification

1. 1Formalize the experimental design problem as a max-potency problem
2. 2Compute the epistemic potency of each potential experiment to measure its worst-case bound reduction
3. 3Select a cost-constrained subset of experiments that maximally tightens bounds on the target query
4. 4Use algorithms and techniques from optimization and causal inference to solve the max-potency problem
5. 5Evaluate the potency of the selected experiments and refine the design as needed

Data scientists and researchers working with causal queries and experimental design can benefit from this knowledge to optimize their experiments and reduce costs

💡 Selecting the right subset of experiments can significantly reduce costs while maximizing the tightening of bounds on target queries

💡 Optimize experiments for causal effect identification with max-potency problem formulation