Doubly Robust Adaptive Conformal Inference for Causal Effects Under Temporal Dependence

The authors introduce a new method called doubly robust adaptive conformal inference (DR-ACI) to create prediction intervals, which are ranges where future values are expected to fall. Their approach works on doubly robust pseudo-outcomes, which combines two techniques to make more reliable predictions. Importantly, this method accounts for data that changes over time, meaning it can handle dependent sequences rather than just independent data.