Bayesian Gaussian Process Modeling for Uncertain Spatial Coordinates Using PyMC

Summary

This article discusses the use of Bayesian spatial probability models, specifically Gaussian processes (GPs), for handling data with uncertain or unknown spatial coordinates. It uses the mining industry as a motivating example, where prospectors take geologic samples to detect mineral resources, but often lack precise location data for those samples. The article explains how PyMC can be used to implement Gaussian process models that account for uncertainty in spatial coordinates, allowing for more robust inference when coordinate data is missing or imprecise.