getconditionals: Get conditional responses from a GP model

In tanyalrogers/GPEDM: Gaussian Process regression for Empirical Dynamic Modeling

Get conditional responses from a GP model

Description

This will obtain responses to each predictor variable across its range of observed values, conditional on all other predictor variables being set to 0 (their mean value, if data are scaled properly). If automatic scaling is used, predictors and responses are backtransformed to their original scale. If the model includes multiple populations, responses are obtained for each population.

Usage

getconditionals(fit, xrange = "default", extrap = 0.01, nvals = 25, plot = T)

Arguments

fit	Output from fitGP.
xrange	The range of scaled predictor values over which responses should be evaluated. Options are "local" (the range observed within each population), or "global" (the range across all populations). Irrelevant if there is only one population. The default ("default") will be "local" if scaling="local" was used, and "global" if scaling="global" or scaling="none" was used. This option can useful for either preventing or allowing for extrapolation beyond the range of the data in a given population.
extrap	Percentage to extrapolate beyond the predictor range for each population (i.e. predictor values will range from (min(X)-range(X)*extrap to (max(X)+range(X)*extrap Defaults to 0.01.
nvals	The number of values to evaluate across the predictor range for each population. Defaults to 25, which should be fine for most applications.
plot	Produce a plot, or not (logical, defaults to TRUE).

Value

Returns (invisibly) a data frame containing the predictor values and conditional responses (means and standard deviations).

tanyalrogers/GPEDM documentation built on June 1, 2025, 8:10 p.m.