residuals: Kolmogorov-Smirnov Test Plots

In andrewzm/INLA: Functions which allow to perform full Bayesian analysis of latent Gaussian models using Integrated Nested Laplace Approximaxion

Description

Illustrate a one-sample Kolmogorov-Smirnov test by plotting the empirical distribution deviation.

Usage

inla.ks.plot(x, y, diff=TRUE, ...)

Arguments

x	a numeric vector of data values.
y	a cumulative distribution function such as 'pnorm'.
diff	logical, indicating if the normalised difference should be plotted. If FALSE, the absolute distribution functions are plotted.
...	additional arguments for ks.test, ignored in the plotting. In particular, only two-sided tests are illustrated.

Details

In addition to the (normalised) empirical distribution deviation, lines for the K-S test statistic are drawn, as well as +/- two standard deviations around the expectation under the null hypothesis.

Value

A list with class "htest", as generated by ks.test

Author(s)

Finn Lindgren [e-mail]

See Also

ks.test

Examples

## Check for N(0,1) data
data = rowSums(matrix(runif(100*12)*2-1,100,12))/2
inla.ks.plot(data, pnorm)

## Not run: 
## Check the goodness-of-fit of cross-validated predictions
result = inla(..., control.predictor=list(cpo=TRUE))
inla.ks.plot(result$pit, punif)

## End(Not run)

andrewzm/INLA documentation built on May 10, 2019, 11:12 a.m.