Bayesian Model Averaging for Data Driven Decision Making when Causality is Partially Known
This addresses the challenge for decision-makers in fields like healthcare or policy where experimentation is infeasible, but it is incremental as it builds on existing causal frameworks with prior knowledge integration.
The paper tackles the problem of making data-driven decisions from observational data when causal relationships are partially known, by developing a method using Bayesian Model Averaging to infer causal graphs and compute expected values and risks of interventions, demonstrating it in various example contexts.
Probabilistic machine learning models are often insufficient to help with decisions on interventions because those models find correlations - not causal relationships. If observational data is only available and experimentation are infeasible, the correct approach to study the impact of an intervention is to invoke Pearl's causality framework. Even that framework assumes that the underlying causal graph is known, which is seldom the case in practice. When the causal structure is not known, one may use out-of-the-box algorithms to find causal dependencies from observational data. However, there exists no method that also accounts for the decision-maker's prior knowledge when developing the causal structure either. The objective of this paper is to develop rational approaches for making decisions from observational data in the presence of causal graph uncertainty and prior knowledge from the decision-maker. We use ensemble methods like Bayesian Model Averaging (BMA) to infer set of causal graphs that can represent the data generation process. We provide decisions by computing the expected value and risk of potential interventions explicitly. We demonstrate our approach by applying them in different example contexts.