Title: High-dimensional Bayesian geostatistics on modest computing environments
Authors: Sudipto Banerjee - UCLA (United States) [presenting]
Abstract: With the growing capabilities of GIS databases, statisticians today routinely encounter geographically referenced data containing observations from a large number of spatial locations and time points. Over the last decade, hierarchical spatio-temporal process models have become widely deployed statistical tools for researchers to better understand the complex nature of spatial and temporal variability. However, fitting such models is often prohibitive in terms of storage and floating point operations with complexity increasing in cubic order for the number of spatial locations and temporal points. This renders such models unfeasible for large data sets. The aim is to discuss model-based strategies to circumvent computational bottlenecks using well-defined massively scalable spatio-temporal stochastic processes. Full Bayesian inference is sought and achieved for millions of locations on very modest computing environments such as R on a standard laptop. These approaches can be described as model-based solutions for big spatio-temporal datasets. The models ensure that the algorithmic complexity has $n$ floating point operations (flops), where $n$ is the number of spatial locations (per iteration). We compare these methods and provide some insight into their methodological underpinnings.