Bayesian structured additive regression modeling of epidemic data: application to cholera

Table 4 Summary of the sensitivity analysis of the choice of hyper-parameters for Model 3

	a = 0.001	a = 0.01	a = 0.5	a = 1
	b = 0.001	b = 0.01	b = 0.0005	b = 0.005
Spatial effects ^‡
$f_{str} (s)$ , $τ_{str}^{2}$	0.02	0.028	0.004	0.004
	(0.0005 - 0.0.06)	(0.003 - 0.07)	(0.00009 - 0.01)	(0.0006 - 0.0009)
$f_{unstr} (s)$ , $τ_{unstr}^{2}$	0.02	0.031	0.007	0.0071
	(0.0009 - 0.0.057)	(0.005 - 0.056)	(0.0001 - 0.028)	(0.0006 - 0.019)
Smooth functions ^§
$f_{1} (ρ_{dump})$ , $τ_{1}^{2}$	0.003	0.006	0.0014	0.002
	(0.0005 - 0.006)	(0.002 - 0.013)	(0.0002 - 0.003)	(0.0006 - 0.004)
$f_{2} (d_{dump})$ , $τ_{2}^{2}$	0.003	0.0078	0.0007	0.002
	(0.0002 - 0.0058)	(0.002 - 0.017)	(0.00008 - 0.0015)	(0.0004 - 0.004)
	0.001	0.004	0.0004	0.001
$f_{3} (d_{reser})$ , $τ_{3}^{2}$	(0.0002 - 0.0024)	(0.001 - 0.009)	(0.00006 - 0.0007)	(0.0004 - 0.003)

‡ Variance components and 90% credible intervals for the spatially structured and unstructured effects; §variance components and 90% credible intervals for the nonlinear smooth functions.

ISSN: 1471-2288