plotObservedEffects: Plot Observed Values Vs. Each Dimension of the Design Matrix

In mlegp: Maximum Likelihood Estimates of Gaussian Processes

Plot Observed Values Vs. Each Dimension of the Design Matrix

Description

Constructs multiple graphs, plotting each parameter from the design matrix on the x-axis and observations on the y-axis

Usage

plotObservedEffects(x, ...)

Arguments

x	an object of class gp or a design matrix
...	if x is a design matrix, a vector of observations; if x is of class gp, a vector of parameter numbers or parameter names to plot (by default, all parameters will be graphed)

Details

if x is NOT of class gp (i.e., x is a design matrix), all columns of x will be plotted separately against the vector of observations

if x is of class gp, the specified columns of the design matrix of x will be plotted against the the observations

Note

It is often useful to use this function before fitting the gaussian process, to check that the observations are valid

Author(s)

Garrett M. Dancik [e-mail]

References

https://github.com/gdancik/mlegp/

Examples

## create the design and output matrices ##
x1 = kronecker(seq(0,1,by=.25), rep(1,5))
x2 = rep(seq(0,1,by=.25),5)
z = 4 * x1 - 2*x2 + x1 * x2 + rnorm(length(x1), sd = 0.001)

## look at the observed effects prior to fitting the GP ##
plotObservedEffects(cbind(x1,x2), z)

## fit the Gaussian process ##
fit = mlegp(cbind(x1,x2), z, param.names = c("x1", "x2"))

## look at the observed effects of the fitted GP (which are same as above)
plotObservedEffects(fit)

mlegp documentation built on March 18, 2022, 5:29 p.m.