Learn R Programming

earth (version 4.1.0)

plot.earth.models: Compare earth models by plotting them.

Description

Compare earth models by plotting them.

Usage

## S3 method for class 'earth.models':
plot(x = stop("no 'x' arg"), which = c(1:2),
    caption = "", jitter = 0,
    col.grsq = discrete.plot.cols(length(x)), lty.grsq = 1,
    col.line = 0, lty.rsq = 5,
    col.vline = col.grsq, lty.vline = 3,
    col.npreds = 0,  lty.npreds  = 2, col.sel.grid = 0,
    ylim = c(0,1),
    col.legend = 1, cex.legend = NULL, legend.pos = NULL, legend.text = NULL,
    col.cum = NULL, do.par = TRUE,
    main = "Model Comparison", cex.main = 1.2, ...)

Arguments

x
A list of one or more earth objects, or a single earth object. This is the only required argument. (This argument is called 'x' for consistency with
which
Which plots to plot: 1 model, 2 cumulative distribution of residuals. Default is 1:2, meaning both.
caption
Overall caption. Values: "string" string "" (default) no caption NULL generate a caption from the $call component of the earth objects.
jitter
Jitter applied to GRSq and RSq values to minimize over-plotting. Default is 0, meaning no jitter. A typical useful value is 0.01. For the col arguments below, 0 means do not plot the corresponding graph element. You ca
col.grsq
Vector of colors for the GRSq plot. The default is discrete.plot.cols(length(x)) which is vector of distinguishable colors, the first three of which are also distinguishable on a monochrome printer. You can examine the colors
lty.grsq
Line type for the GRSq plot. Default is 1.
col.line
Vector of colors for the RSq plot. Default is 0, meaning no RSq plot.
lty.rsq
Line type for the RSq plot. Default is 5.
col.vline
A vertical line is drawn for each object to show which model size was chosen for that object. The color of the line is col.vline. Default is col.grsq.
lty.vline
Line type of vertical lines (a vertical line is drawn to show the selected model for each object). Can be a vector. Default is 3.
col.npreds
Vector of colors for the "number of predictors" plot within the model selection plot. Default is 0, meaning no "number of predictors" plot. The special value NULL means borrow col.grsq (or col.line<
lty.npreds
Line type of the "number of predictors" plot (in the Model Selection plot). Default is 2.
col.sel.grid
Color of grid lines in the Model Selection graph. Default is 0, no grid. Try something like "lightgray", "linen", or "seashell".
ylim
Two element vector c(min,max) specifying min and max values on the y axis of the RSq/GRSq plot. Default is c(0,1). The special value min=-1 means the minimum y axis value is the smallest GRSq or RSq, excl
col.legend
Deprecated. Use legend.pos=NA for no legend.
cex.legend
Legend cex. Default is NULL, meaning choose automatically.
legend.pos
Legend position. Default NULL, meaning position the legend automatically. Use NA for no legend. Specify c(x,y) in user coordinates, or use "topleft" etc. as explained in
legend.text
Vector of strings to use as legend text. The special value NULL (default) means generate the legend text automatically from call$formula.
col.cum
Vector of colors for the cumulative distribution plot. The special value NULL (default) means borrow col.grsq (or col.line if col.grsq is NULL). The following settings are rela
do.par
Call par() for global settings as appropriate. Default is TRUE, which sets mfrow, mar=c(4,4,2,3), mgp=c(1.6,0.6,0), cex=0.7. Set to FALSE if you want to append figures to an existing plot.
main
Title of each plot. Default is NULL, meaning generate figure headings automatically.
cex.main
Size of plot titles. Default is 1.2. Used only if do.par is TRUE (default).
...
Unused, but provided for generic/method consistency.

See Also

earth, plot.earth, plot.earth.models, plotd, plotmo

Examples

Run this code
data(ozone1)
a1 <- earth(O3 ~ .,          data = ozone1, degree = 2)
a2 <- earth(O3 ~ .-wind,     data = ozone1, degree = 2)
a3 <- earth(O3 ~ .-humidity, data = ozone1, degree = 2)
plot.earth.models(list(a1,a2,a3), ylim=c(.65,.85))

Run the code above in your browser using DataLab