Gaussian Process Regression Tuned by Bayesian Optimization for Seawater Intrusion Prediction

Table 2

Value ranges of optimized hyperparameters of regressors.

Regressor name	Parameter(s) name(s)	Observed value range (number of points appeared)
GPR	Kernel function	Squared exponential (30/30)
	Sigma	0.013 ± 0.005 (26/30)
LRM	Learner	Least squares
	Initial bias	−0.85 ± 0.04 (30/30)
SVM	Kernel function	Polynomial (8/20)
		Gaussian RBF (11/30)
		Linear (11/30)
BRDT	Max splits	500 ± 200 (8/30)
		900 ± 200 (8/30)
		1300 ± 200 (3/30)
		≥1501 (11/30)
	Number of variables to sample	All
TreeEns	Number of learners	200 ± 200 (17/30)
		≥400 (13/30)