Online Program Home
My Program

Abstract Details

Activity Number: 429
Type: Contributed
Date/Time: Tuesday, August 2, 2016 : 2:00 PM to 3:50 PM
Sponsor: Section on Statistics and the Environment
Abstract #320126 View Presentation
Title: Spatial Prediction of Naturally Occurring Indoor Gamma Radiation Dose Rates in Great Britain
Author(s): Mark Peter Little* and Pavel Chernyavskiy and Gerald M. Kendall and Philip S. Rosenberg and Richard Wakeford
Companies: National Cancer Institute and National Cancer Institute and University of Oxford and National Cancer Institute and University of Manchester
Keywords: ionizing radiation ; childhood leukemia ; childhood cancer ; multi-resolution Gaussian process ; geostatistical models ; natural background gamma radiation
Abstract:

Gamma radiation is an important part of background radiation, and correlates with childhood leukemia risk in Great Britain (GB). The spatial variation of indoor gamma radiation dose-rates in GB is explored using various geostatistical methods. A multi-resolution Gaussian process (MRGP) model using radial basis functions with 2, 4, or 8 components, is fitted via maximum likelihood, and a non-spatial model is also used, fitted by ordinary least squares (OLS); because of the dataset size (N=10,199), four other spatial models are fitted by variogram-fitting. A randomly selected 70:30 fitting:validation split is used. The models are evaluated based on their Mean Absolute Error, Mean Squared Error, as well as Pearson/rank correlation between predicted and actual dose-rates. Each of the Matérn, Gaussian, Bessel, and Spherical models fitted the empirical variogram well, yielding similar predictions at >50 km separation, although with greater differences in predicted variograms at < 50 km. The 8-component MRGP model had the best predictive accuracy. The Spherical, Bessel, Matérn, Gaussian and OLS models had progressively worse predictive performance, the OLS model being particularly poor.


Authors who are presenting talks have a * after their name.

Back to the full JSM 2016 program

 
 
Copyright © American Statistical Association