Using confounders to model uncertainty in occluded scenarios

How do you model ”what you don’t know” to give you insights into when to avoid an unwanted experience? Can causal analytics be a mathematical approach to modeling unidentified bias to determine how to avoid safety hazard scenarios and provide alternative solutions? Unidentified confounders are used to address emergent occluded scenarios reflective of uncertainty during deployed operations. DAGs and do-calculus equations applied to the backdoor adjustment formula analyze various degrees of performance degradation. These equations model the effects of occluded scenarios as counterfactuals identifying performance confounder bias. From confounder bias, performance degradation can reach unacceptable thresholds, providing an argument for proactive adaptability training in the form of trained fallback policies based on uncertainty levels to avoid unwanted experiences.