GeDS (version 0.2.5)

plot,GeDS-method: Plot method for GeDS objects.

Description

Plot method for GeDS objects. Plots GeDS fits.

Usage

# S4 method for GeDS,ANY
plot(
  x,
  f = NULL,
  which,
  DEV = FALSE,
  ask = FALSE,
  main,
  legend.pos = "topright",
  new.window = FALSE,
  wait = 0.5,
  n = 3L,
  type = c("none", "Polygon", "NCI", "ACI"),
  ...
)

Arguments

x

a GeDS-class object from which the GeDS fit(s) should be extracted.

f

(optional) specifies the underlying function or generating process to which the model was fit. This parameter is useful if the user wishes to plot the specified function/process alongside the model fit and the data

which

a numeric vector specifying the iterations of stage A for which the corresponding GeDS fits should be plotted. It has to be a subset of 1:nrow(x$stored). See details.

DEV

logical variable specifying whether a plot representing the deviance at each iteration of stage A should be produced or not.

ask

logical variable specifying whether the user should be prompted before changing the plot page.

main

optional character string to be used as a title of the plot.

legend.pos

the position of the legend within the panel. See legend for details.

new.window

logical variable specifying whether the plot should be shown in a new window or in the active one.

wait

time, in seconds, the system should wait before plotting a new page. Ignored if ask = TRUE.

n

integer value (2, 3 or 4) specifying the order (\(=\) degree \(+ 1\)) of the GeDS fit that should be plotted. By default equal to 3L. Non-integer values will be passed to the function as.integer.

type

character string specifying the type of plot required. Should be set either to "Polygon" if the user wants to get also the control polygon of the GeDS fit, "NCI" or "ACI" if 95% confidence bands for the predictions should be plotted (see details) or "none" if only the fitted GeDS curve should be plotted. Applies only when plotting a univariate spline regression.

...

further arguments to be passed to the plot.default function.

Details

This method is provided in order to allow the user to plot the GeDS fits contained in the GeDS-Class objects.

Since in Stage A of the GeDS algorithm the knots of a linear spline fit are sequentially located, one at a time, the user may wish to visually inspect this process using the argument which. The latter specifies a particular iteration number (or a vector of such numbers) for which the corresponding linear fit(s) should be plotted. The ask and wait arguments can help the user to manage these pages.

By means of ask the user can determine for how long each page should appear on the screen. Pages are sequentially replaced by pressing the enter button.

Note that, in order to ensure stability, if the object was produced by the function GGeDS, plotting intermediate fits of stage A is allowed only if n = 2, in contrast to objects produced by NGeDS for which plotting intermediate results is allowed also for n = 2 or 3 results.

The confidence intervals obtained by setting type = "NCI" are approximate local bands obtained considering the knots as fixed constants. Hence the columns of the design matrix are seen as covariates and standard methodology relying on the se.fit option of predict.lm or predict.glm is applied.

Setting type = "ACI", asymptotic confidence intervals are plotted. This option is applicable only if the canonical link function has been used in the fitting procedure.

See Also

NGeDS and GGeDS; plot.

Examples

###################################################
# Generate a data sample for the response variable
# Y and the single covariate X, assuming Normal noise
set.seed(123)
N <- 500
f_1 <- function(x) (10*x/(1+100*x^2))*4+4
X <- sort(runif(N, min = -2, max = 2))
# Specify a model for the mean of Y to include only a component
# non-linear in X, defined by the function f_1
means <- f_1(X)
# Add (Normal) noise to the mean of Y
Y <- rnorm(N, means, sd = 0.1)

# Fit a Normal GeDS regression using NGeDS
(Gmod <- NGeDS(Y ~ f(X), beta = 0.6, phi = 0.995, Xextr = c(-2,2)))

# Plot the final quadratic GeDS fit (red solid line)
# with its control polygon (blue dashed line)
plot(Gmod)

# Plot the quadratic fit obtained from the linear fit at the 10th
# iteration of stage A i.e. after 9 internal knots have been inserted
# by the GeDS procedure
plot(Gmod, which=10)

# Generate plots of all the intermediate fits obtained
# by running the GeDS procedure
if (FALSE) {
plot(Gmod, which=1:16)
}

###################################################
# Generate a data sample for the response variable Y and the covariate
# X assuming Poisson distributed error and a log link function

set.seed(123)
N <- 500
f_1 <- function(x) (10*x/(1+100*x^2))*4+4
X <- sort(runif(N ,min = -2, max = 2))
# Specify a model for the mean of Y to include only a component
# non-linear in X, defined by the function f_1
means <- exp(f_1(X))
# Generate Poisson distributed Y according to the mean model
Y <- rpois(N,means)

# Fit a Poisson GeDS regression model using GGeDS
(Gmod2 <- GGeDS(Y ~ f(X), beta = 0.2, phi = 0.995, family = poisson(),
                Xextr = c(-2,2)))

# similar plots as before, but for the linear fit
plot(Gmod2, n = 2)
plot(Gmod2, which = 10, n = 2)
if (FALSE) {
plot(Gmod2, which = 1:16, n = 2)
plot(Gmod2, which = 1:16, n = 2, ask = T)
}