ks (version 1.5.2)

plot.kda.kde: Kernel discriminant analysis plot for 1- to 3-dimensional data

Description

Kernel discriminant analysis plot for 1- to 3-dimensional data.

Usage

## univariate
## S3 method for class 'kda.kde':
plot(x, y, y.group, prior.prob=NULL, xlim, ylim,
    xlab="x", ylab="Weighted density function", drawpoints=TRUE,
    lty, lcol, col, ptcol, ...)
## bivariate
## S3 method for class 'kda.kde':
plot(x, y, y.group, prior.prob=NULL, cont=c(25,50,75),
    abs.cont, xlim, ylim, xlab, ylab, drawpoints=FALSE,
    drawlabels=TRUE, cex=1, pch, lty, col, lcol, ptcol, ...)
## trivariate
## S3 method for class 'kda.kde':
plot(x, y, y.group, prior.prob=NULL, cont=c(25,50), colors,
   alphavec, xlab, ylab, zlab, drawpoints=FALSE, size=3,
   ptcol="blue", ...)

Arguments

x

an object of class kda.kde (output from kda.kde)

y

matrix of test data points

y.group

vector of group labels for test data points

prior.prob

vector of prior probabilities

cont

vector of percentiles of density estimate heights for contour level curves

abs.cont

vector of absolute density estimate heights for contour level curves - only one of cont or abs.cont needs to be specified

cex,pch,lty,xlim,ylim,xlab,ylab,zlab

usual graphics parameters

drawpoints

if TRUE then draw data points

drawlabels

if TRUE then draw contour labels (2-d plot)

col

vector of colours for partition classes

ptcol

vector of colours for data points of each group

lcol

vector of colours for contour lines of density estimates

colors

vector of colours for contours of density estimates (3-d plot)

alphavec

vector of transparency values - one for each contour (3-d plot)

size

size of plotting symbol (3-d plot)

...

other graphics parameters

Value

• Plot of 1-d and 2-d density estimates for discriminant analysis is sent to graphics window. Plot for 3-d is sent to RGL window.

synopsis

## S3 method for class 'kda.kde': plot(x, y, y.group, ...)

Details

If prior.prob is set to a particular value then this is used. The default is NULL which means that the sample proportions are used.

The object x contains the training data and its group labels. If y and y.group are missing then the training data points are plotted. Otherwise, the test data y are plotted.

-- For 1-d plots: The partition induced by the discriminant analysis is plotted as rug plot (with the ticks inside the axes). If drawpoints=TRUE then the data points are plotted as a rug plot with the ticks outside the axes, their colour is controlled by ptcol. -- For 2-d plots: If display="part" then a partition induced by the discriminant analysis is also plotted. If this is not desired, set display="". Its colours are controlled by col (the default is heat.colors). The plotting symbols are set by pch and the colour by ptcol. The default contours are at 25%, 50%, 75% or cont=c(25,50,75) for the highest density regions, see Hyndman (1996). The line types are set by lty.

-- For 3-d plots: Default contours are cont=c(25,50) for the highest density regions, see Hyndman (1996). The colour for each group is set using colors - default is heat.colors. The transparency of each contour (within each group) is alphavec, with the default range from 0.1 to 0.5.

References

Bowman, A.W. & Azzalini, A. (1997) Applied Smoothing Techniques for Data Analysis. Clarendon Press. Oxford.

Hyndman, R.J. (1996) Computing and graphing highest density regions, American Statistician, 50, 120--126

Simonoff, J. S., (1996) Smoothing Methods in Statistics. Springer-Verlag. New York.

See Also

kda.kde, kda

Examples

library(MASS)
data(iris)

xlim <- c(4,8)
ylim <- c(2,4.5)

## univariate example
ir <- iris[,1]
ir.gr <- iris[,5]

kda.fhat <- kda.kde(ir, ir.gr, hs=sqrt(c(0.01, 0.04, 0.07)))
plot(kda.fhat, xlab="Sepal length", ptcol=1:3)

## bivariate example
ir <- iris[,1:2]
ir.gr <- iris[,5]
H <- Hkda(ir, ir.gr, bw="plugin", pre="scale")
kda.fhat <- kda.kde(ir, ir.gr, Hs=H)

plot(kda.fhat, cont=0)
plot(kda.fhat, xlim=xlim, ylim=ylim, col=c("transparent", "grey", "#8f8f8f"))

## trivariate example

ir <- iris[,1:3]
ir.gr <- iris[,5] 
H <- Hkda(ir, ir.gr, bw="plugin", pre="scale")
kda.fhat <- kda.kde(ir, ir.gr, Hs=H)
plot(kda.fhat)