Display of Fitted LOESS Models by Coplots
DESCRIPTION:
Plots or prepares to plot an object of class "loess".
USAGE: 
plot.loess(object, given=6, evaluation=50, confidence=0,
     coverage=0.99, ranges, which.plots)


preplot.loess(object, given=6, evaluation=50, confidence=0,
     coverage=0.99, ranges, which.plots)
REQUIRED ARGUMENTS:
object:
a loess object.
OPTIONAL ARGUMENTS:
given:
number of conditioning values for a numeric given variable.
evaluation:
number of points at which the curve on each dependence panel is computed.
confidence:
number of points at which the confidence intervals on each dependence panel are displayed.
coverage:
the level of the confidence intervals expressed as a fraction.
ranges:
the ranges of evaluations of numeric variables. If there are k numeric predictors, ranges should either be a list of minimum and maximum pairs of length k, or a 2 by k matrix. Matching is done the same way as for xlab. If missing, the range for a numeric predictor is the range of the predictor observations used in the fit.
which.plots:
a character vector of names of predictors that determines the coplots that are made. If missing, all coplots are made.

Graphical parameters and arguments to title may also be given to plot.loess.

VALUE:
preplot.loess returns an object of class "preplot.loess" with the evaluations necessary to draw coplots using plot.preplot.loess, which is a method for plot.

plot.loess returns NULL.

SIDE EFFECTS:
plot.loess creates a plot on the current graphics device.
DETAILS:
This function is a method for the generic function plot for class "loess". It can be invoked by calling plot for an object of the appropriate class, or directly by calling plot.loess regardless of the class of the object.

Sometimes, the computation necessary to carry out plot.loess is substantial enough that we want to save the evaluated surface that is graphed so we can graph the surface many times, perhaps with changes to graphical parameters, but carry out only one evaluation. This can be done by assigning the result of preplot to a dataset and using the dataset as the first argument in calls to plot.

BUGS:
plot.loess ignores ask=T. You must set ask=T with par before calling plot.loess.

SEE ALSO:
plot , plot.preplot.loess , loess , predict.loess , coplot .
EXAMPLES: 
ethanol.model <- loess(NOx ~ C * E, data = ethanol, span = 1/2,
      drop.square = "C", parametric = "C")
ethanol.preplot <- preplot(ethanol.model, given = 16, confidence = 7,
      coverage = .95)
plot(ethanol.preplot)
# now try again with a smaller character size
plot(ethanol.preplot, cex = .5)
# now make hard copy
postscript()
plot(ethanol.preplot, cex = .5)