Overview
The study proposes Doubly Robust Adaptive Conformal Inference (DR-ACI), a methodology designed for the construction of prediction intervals. This method specifically targets doubly robust pseudo-outcomes within contexts characterized by temporal dependence.
Research Context
The core problem addressed is the generation of prediction intervals for outcomes under conditions where temporal dependence is a significant factor. The method utilizes a doubly robust framework for pseudo-outcome estimation.
Approach
The proposed approach, DR-ACI, integrates principles of adaptive conformal inference with doubly robust estimation techniques. Its application is framed around pseudo-outcomes, which are constructs often used in causal inference. The crucial element of the approach is its explicit consideration and handling of temporal dependence. The inference procedure is adapted to account for the characteristics of data exhibiting temporal correlations.
Findings
The research presents DR-ACI as a method that constructs prediction intervals. These intervals are specifically for doubly robust pseudo-outcomes. A key characteristic of the method is its ability to operate effectively under conditions of temporal dependence.