SLGP (version 1.0.2)

predictSLGP_moments: Predict centered or uncentered moments at new locations from a SLGP model

Description

Computes statistical moments (e.g., mean, variance, ...) of the posterior predictive distributions at new covariate locations, using a given SLGP model.

Usage

predictSLGP_moments(
  SLGPmodel,
  newNodes,
  power,
  centered = FALSE,
  interpolateBasisFun = "WNN",
  nIntegral = 101,
  nDiscret = 101,
  discrete = FALSE
)

Value

A data frame with:

• Repeated rows of the input covariates,

• A column power indicating the moment order,

• One or more columns mSLGP_1, mSLGP_2, ... for the estimated moments across posterior samples.

Arguments

SLGPmodel

An object of class SLGP-class.

newNodes

A data frame of new covariate values.

power

Scalar or vector of positive integers indicating the moment orders to compute.

centered

Logical; if TRUE, computes centered moments. If FALSE, computes raw moments.

interpolateBasisFun

Interpolation mode for basis functions: "nothing", "NN", or "WNN" (default).

nIntegral

Number of integration points for computing densities.

nDiscret

Discretization resolution of the response space.

discrete

Boolean, indicates if we work with continuous pdfs (default, FALSE) or discrete probabilities

Examples

# \donttest{
# Load Boston housing dataset
library(MASS)
data("Boston")
# Set input and output ranges manually (you can also use range(Boston$age), etc.)
range_x <- c(0, 100)
range_response <- c(0, 50)

# Train an SLGP model using Laplace estimation and RFF basis
modelLaplace <- slgp(medv ~ age,        # Use a formula to specify response and covariates
                 data = Boston,     # Use the original Boston housing data
                 method = "Laplace",    # Train using Maximum A Posteriori estimation
                 basisFunctionsUsed = "RFF",         # Random Fourier Features
                 sigmaEstimationMethod = "heuristic",  # Auto-tune sigma2 (more stable)
                 predictorsLower = range_x[1],         # Lower bound for 'age'
                 predictorsUpper = range_x[2],         # Upper bound for 'age'
                 responseRange = range_response,       # Range for 'medv'
                 opts_BasisFun = list(nFreq = 200,     # Use 200 Fourier features
                                      MatParam = 5/2), # Matern 5/2 kernel
                 seed = 1)                             # Reproducibility
dfX <- data.frame(age=seq(range_x[1], range_x[2], 1))
predMean <- predictSLGP_moments(SLGPmodel=modelLaplace,
                                newNodes = dfX,
                                power=c(1, 2),
                                centered=FALSE) # Uncentered moments of order 1 and 2
predVar <- predictSLGP_moments(SLGPmodel=modelLaplace,
                               newNodes = dfX,
                               power=c(2),
                               centered=TRUE) # Centered moments of order 2 (Variance)
# }