ABSTRACT. We have access to an unprecedented amount of fine-grained data on cities, transportation, economies, and societies, much of these data referenced in geo-space and time. There is a tremendous opportunity to discover new knowledge about spatial economies that can inform theory and modeling in regional science. However, there is little evidence of computational methods for discovering knowledge from databases in the regional science literature. This paper addresses this gap by clarifying the geospatial knowledge discovery process, its relation to scientific knowledge construction, and identifying challenges to a greater role in regional science.