GeoShapley: A Game Theory Approach to Measuring Spatial Effects in Machine Learning Models

arXiv:2312.03675v2 Announce Type: replace
Abstract: This paper introduces GeoShapley, a game theory approach to measuring spatial effects in machine learning models. GeoShapley extends the Nobel Prize-winning Shapley value framework in game theory by conceptualizing location as a player in a model prediction game, which enables the quantification of the importance of location and the synergies between location and other features in a model. GeoShapley is a model-agnostic approach and can be applied to statistical or black-box machine learning models in …

abstract arxiv cs.lg effects framework game game theory location machine machine learning machine learning models measuring paper prediction prize quantification spatial stat.ml theory type value