Efficient Data-Driven Knot Selection for Reduced Rank Spatial Models

• Activity Number: 85
• Type: Contributed
• Date/Time: Sunday, August 3, 2014 : 4:00 PM to 5:50 AM
• Sponsor: Section on Statistics and the Environment
• Author(s): Casey M. Jelsema*+ and Shyamal Peddada
• Companies: National Institute of Environmental Health Sciences and NIH/NIEHS
• Keywords: big data ; spatial mixed effects ; knot selection ; reduced rank spatial models ; GuLF Study

Abstract

Modern advances have enabled the collection of massive spatial and spatio-temporal datasets (upwards of tens of thousands of observations). Reduced rank models are typically required for computational feasibility when analyzing such datasets. Multiple strategies exist for specifying reduced rank models (RRMs) for both spatial and spatio-temporal models. However, a common feature is to define a reduced-dimension latent process over a selected number of knot locations across the spatial domain. Most work on RRMs focuses on estimation of parameters or specification of appropriate prior distributions. But the issue of selecting the knot locations has been receiving increasing attention. Current methods for knot selection often rely on Bayesian methods, which are potentially computationally intensive. In this paper we propose an efficient, data-driven approach to knot selection for reduced rank spatial models. We illustrate our method through simulations and on a real-world dataset.