Skip to main content

Table 3 Performance comparison of Bayesian and geostatistical estimators: mean absolute error of prediction.

From: How does Poisson kriging compare to the popular BYM model for mapping disease risks?

Estimators Lung cancer (mean = 21.25) Cervix cancer (mean = 2.993)
Arithmetical average Average % best result Average % best result
Global Empirical Bayes 1.396 2 0.517 0
Local Empirical Bayes 1.380 0 0.458 0
BYM model 1.280 6 0.426 0
Point Poisson kriging (adjacent counties) 1.243 30 0.400 16
ATA Poisson kriging (adjacent counties) 1.246 32 0.397 14
ATA Poisson kriging (32 neighbors) 1.250 30 0.380 70
Population-weighted average     
Global Empirical Bayes 0.972 2 0.152 0
Local Empirical Bayes 0.980 2 0.145 0
BYM model 0.918 12 0.141 4
Point Poisson kriging (adjacent counties) 0.903 32 0.133 18
ATA Poisson kriging (adjacent counties) 0.909 24 0.132 28
ATA Poisson kriging (32 neighbors) 0.911 28 0.132 50
  1. Results obtained on average (arithmetical and population-weighted) over 50 realizations generated for Regions 1 and 2. Poisson kriging was conducted using either adjacent counties (same neighbors as BYM model) or the 32 closest counties in terms of distance between population-weighted centroids. ATA kriging accounts for the shape and size of the counties in the analysis. Straightforward empirical Bayesian smoothers were also applied. Bold numbers refer to best performances. The second column gives the percentage of realizations where the particular method yields the smallest prediction error.