Skip to main content
. Author manuscript; available in PMC: 2020 Aug 2.
Published in final edited form as: Med Decis Making. 2019 Aug 2;39(5):540–552. doi: 10.1177/0272989X19862560

Table 3.

Hyperparameter settings for Gaussian process regressors from Bayesian optimization, r=j=119(θjθj)2σj2 is weighted Euclidean distance between parameters θ and θ′.

Regressor Type of Basis Function Kernel Function
y^1 Linear k(θ,θ)=σf2(1+5r+53r2)exp(5r),σf=0.2091
y^2 Constant k(θ,θ)=σf2(1+5r+53r2)exp(5r),σf=0.3101
y^3 Constant k(θ,θ)=σf2(1+r22α)α,σf=0.4558,α=0.063