Predict: The Prediction Function of 'GPM' Package
In GPM: Gaussian Process Modeling of Multi-Response and Possibly Noisy Datasets

Predicts the reponse(s), associated prediction uncertainties, and gradient(s) of the GP model fitted by Fit.

1	Predict(XF, Model, MSE_on = 0, YgF_on = 0, grad_dim = rep(1, ncol(XF)))

`XF`	Matrix containing the locations (settings) where the predictions are desired. The rows and columns of `XF` denote individual observation settings and input dimension, respectively.
`Model`	The GP model fitted by `Fit`.
`MSE_on`	Flag (a scalar) indicating whether the uncertainty (i.e., mean squared error `MSE`) associated with prediction of the response(s) should be calculated. Set to a non-zero value to calculate `MSE`.
`YgF_on`	Flag (a scalar) indicating whether the gradient(s) of the response(s) are desired. Set to a non-zero value to calculate the gradient(s). See `note` below.
`grad_dim`	A binary vector of length `ncol(XF)`. The gradient of the response(s) will be calculated along the dimensions where the corresponding element of `grad_dim` is `1`. `grad_dim` is ignored if `YgF_on == 0`.

Output A list containing the following components:

YF A matrix with n rows (the number of prediction points) and dy columns (the number of responses).
MSE A matrix with n rows and dy columns where each element represents the prediction uncertainty (i.e., the expected value of the squared difference between the prediction and the true response) associated with the corresponding element in YF.
YgF An array of size n by sum{grad_dim} by dx.

The gradient(s) can be calculated if CorrType='G' or CorrType='LBG'. If CorrType='PE' or CorrType='LB', the gradient(s) can only be calculated if Power = 2 and Gamma = 1, respectively.
For efficiency, make sure the inputs are vecotrized and then passed to Predict. Avoid passing inputs individually in a for loop.

Bostanabad, R., Kearney, T., Tao, S., Apley, D. W. & Chen, W. (2018) Leveraging the nugget parameter for efficient Gaussian process modeling. Int J Numer Meth Eng, 114, 501-516.
Plumlee, M. & Apley, D. W. (2017) Lifted Brownian kriging models. Technometrics, 59, 165-177.

Fit to see how a GP model can be fitted to a training dataset.
Draw to plot the response via the fitted model.