GeoConformal Prediction: A Model-Agnostic Framework for Measuring the Uncertainty of Spatial Prediction

Abstract

Spatial prediction is a fundamental task in geography, providing essential data support for various scenarios.Recent advancements, empowered by the development of geospatial artificial intelligence (GeoAI), haveprimarily focused on improving prediction accuracy while overlooking reliable measurements of predictionuncertainty. Such measures are crucial for enhancing model trustworthiness and supporting responsibledecision-making. To address this issue, we propose a model-agnostic uncertainty assessment method calledGeoConformal Prediction (GeoCP). First, a simulation study is conducted to validate the usefulness ofGeoCP. Then, we applied GeoCP to two classic spatial prediction cases, spatial regression and spatialinterpolation, to evaluate its reliability. For the case of spatial regression, we used XGBoost to predicthousing prices, followed by GeoCP to calculate uncertainty. Our results show that GeoCP achieved acoverage rate of 93.67 percent, whereas bootstrapping methods reached a maximum coverage of 81.00percent after 2,000 runs. We then applied GeoCP for the case of spatial interpolation models. By comparinga GeoAI-based geostatistical model with a traditional geostatistical model (Kriging), we found that theuncertainty obtained from GeoCP aligned closely with the variance in Kriging. Finally, using GeoCP, weanalyzed the sources of uncertainty in spatial prediction. We found that explicitly including local features inAI models can significantly reduce prediction uncertainty, especially in areas with strong local dependence.Our findings suggest that GeoCP holds substantial potential not only for geographic knowledge discovery butalso for guiding the design of future GeoAI models, paving the way for more reliable and interpretablespatial prediction frameworks. The method is implemented in an open-source Python package namedgeoconformal. Key Words: conformal prediction, GeoAI, Kriging, spatial regression, spatial uncertainty.