plotmvoutlier function

Multivariate outlier plot

This function plots multivariate outliers. One possibility is to distinguish between outlier and no outlier. The alternative is to distinguish between the different percentils (e.g. <25%, 25%<x<50%,...).

plotmvoutlier(coord, data, quan = 1/2, alpha = 0.025, symb = FALSE, bw = FALSE,
plotmap = TRUE, map = "kola.background", which.map = c(1, 2, 3, 4),
map.col = c(5, 1, 3, 4), map.lwd = c(2, 1, 2, 1), pch2 = c(3, 21),
cex2 = c(0.7, 0.2), col2 = c(1, 1), lcex.fac = 1, ...)

Arguments

• coord: the coordinates for the points
• data: the value for the different coordinates
• quan: Number of subsets used for the robust estimation of the covariance matrix. Allowed are values between 0.5 and 1., see covMcd
• alpha: Maximum thresholding proportion
• symb: if FALSE, only two different symbols (outlier and no outlier) will be used
• bw: if TRUE, symbols are in gray-scale (only if symb=TRUE)
• plotmap: if TRUE, the map is plotted
• map: the name of the background map
• which.map, map.col, map.lwd: parameters for the background plot, see plotbg
• pch2, cex2, col2: graphical parameters for the points
• lcex.fac: factor for multiplication of symbol size (only if symb=TRUE)
• ...: further parameters for the plot

Details

The function computes a robust estimation of the covariance and then the Mahalanobis distances are calculated. With this distances the data set is divided into outliers and non outliers. If symb=FALSE only two different symbols are used otherwise different grey scales are used to distinguish the different types of outliers.

Returns

• o: returns the outliers

• md: the square root of the Mahalanobis distance

• euclidean: the Euclidean distance of the scaled data

References

C. Reimann, P. Filzmoser, R.G. Garrett, and R. Dutter: Statistical Data Analysis Explained. Applied Environmental Statistics with R. John Wiley and Sons, Chichester, 2008.

Author(s)

Peter Filzmoser <P.Filzmoser@tuwien.ac.at > http://cstat.tuwien.ac.at/filz/

See Also

plotbg, covMcd, arw

Examples

data(moss)
X=moss[,"XCOO"]
Y=moss[,"YCOO"]
el=c("Ag","As","Bi","Cd","Co","Cu","Ni")
x=log10(moss[,el])

data(kola.background)
plotmvoutlier(cbind(X,Y),x,symb=FALSE,map.col=c("grey","grey","grey","grey"),
       map.lwd=c(1,1,1,1),
       xlab="",ylab="",frame.plot=FALSE,xaxt="n",yaxt="n")