Search
Menu
All issues Volume 3 (2020) 4open, 3 (2020) 12 Full HTML
Open Access

Table 3

Summaries of the Bayesian estimation.

prior mean s–m sd 2.5% 25% 50% 75% 97.5% n_eff Rhat
Case VT vs. VF – see Table 1
Gaussian NI 0.856 0.001 0.043 0.752 0.834 0.863 0.886 0.916 1161 1.000
Gaussian I 0.859 0.001 0.040 0.765 0.838 0.865 0.887 0.918 1198 1.003
Gumbel H NI 3.023 0.012 0.445 2.243 2.707 3.001 3.325 3.938 1315 1.003
Gumbel H I 3.021 0.012 0.444 2.222 2.712 3.005 3.298 3.931 1320 1.001
H × VF vs. Dmax – see Table 2 (left)
Gaussian NI 0.696 0.003 0.088 0.490 0.648 0.712 0.760 0.822 1167 1.002
Gaussian I 0.707 0.002 0.082 0.506 0.663 0.722 0.765 0.824 1305 1.000
Gumbel H NI 2.100 0.009 0.317 1.512 1.881 2.089 2.302 2.743 1158 1.003
Gumbel H I 2.111 0.009 0.319 1.540 1.883 2.089 2.317 2.805 1222 1.001
Hf vs. Dmax – see Table 2 (right)
Gaussian NI 0.768 0.002 0.067 0.612 0.732 0.779 0.817 0.864 1286 1.002
Gaussian I 0.771 0.002 0.065 0.618 0.736 0.782 0.819 0.864 1192 1.003
Gumbel H NI 2.392 0.010 0.358 1.746 2.135 2.385 2.629 3.120 1273 1.001
Gumbel H I 2.399 0.012 0.362 1.742 2.144 2.372 2.642 3.135 959 1.001

n_eff: final number of simulations used for the estimation; sd: standard deviation; s–m = sd/n_eff1/2; Rhat: potential scale reduction factor on split chains (at convergence, Rhat = 1). In bold, the Bayesian estimates of ρ for Gaussian copula and θ for Gumbel–Hougaard copula, by quadratic loss function on left, by multilinear loss function on right. On the top, Non-Informative (NI) prior; on the bottom, Informative (I).