ABSTRACT
Spatial prediction methods represent a set of tools for obtaining accurate data of geographic variables from limited observations. As an emerging subfield of GIScience that uses artificial intelligence and machine learning techniques for geographic knowledge discovery, GeoAI offers a novel and bold perspective on revisiting and improving current spatial prediction and interpolation methods. In this chapter, the GeoAI motivations of spatial data representation, spatial structure measuring and the spatial relationship modeling throughout the workflow of spatial prediction are presented in the context of leveraging AI techniques. This chapter reviewed GeoAI for spatial prediction and interpolation methods, with a particular focus on two major fields: geostatistics and spatial regression. Challenges are discussed around uncertainty, transferability and interpretability.
