Conformal Prediction for Functional Kriging Models

In this work we introduce a conformal prediction method for functional kriging. Conformal Prediction (CP) is a framework in machine learning and statistical inference that provides a principled way to quantify uncertainty and make predictions without relying on specific distributional assumptions. The approach we introduce provides additional information about the uncertainty associated with the kriging predictions by constructing prediction regions with a specified error rate. We apply CP for functional kriging to the analysis of spatially located network information. This allows us to obtain more reliable estimates and better assess the accuracy of predictions based on functional data.