Causal Classification for Decision-Making with Continuous Outcomes

Causal decision-making uses causal information derived from data to determine whether a treatment or intervention should be implemented. Most existing approaches inform such decisions by estimating conditional average treatment effects (CATEs), which quantify average treatment effects within subgroups defined by observed characteristics. More recently, causal classification has emerged as an alternative framework for causal decision-making that directly predicts whether an intervention has a positive causal effect for a given individual. However, existing research has largely focused on binary outcomes. We extend this framework to continuous outcomes by predicting the sign of the individual causal effect (ICE).