heplots (version 1.2-0)

cqplot: Chi Square Quantile-Quantile plots

Description

A chi square quantile-quantile plots show the relationship between data-based values which should be distributed as $\chi^2$ and corresponding quantiles from the $\chi^2$ distribution. In multivariate analyses, this is often to assess multivariate normality and check for outliers, using the Mahalanobis squared distances ($D^2$) of observations from the centroid. cqplot is a more general version of similar functions in other packages that produce chi square QQ plots. It allows for classical Mahalanobis squared distances as well as robust estimates based on the MVE and MCD; it provides an approximate confidence (concentration) envelope around the line of unit slope, a detrended version, where the reference line is horizontal, the ability to identify or label unusual points, and other graphical features. The method for "mlm" objects applies this to the residuals from the model.

Usage

cqplot(x, ...)

## S3 method for class 'mlm':
cqplot(x, ...)

## S3 method for class 'default':
cqplot(x, method = c("classical", "mcd", "mve"), 
	detrend = FALSE, pch = 19, col = palette()[1], 
	ref.col = "red", ref.lwd = 2, 
	conf = 0.95, env.col = "gray", env.lwd = 2, env.lty = 1, 
	env.fill = TRUE, fill.alpha = 0.2, 
	fill.color = trans.colors(ref.col, fill.alpha), 
	labels = if (!is.null(rownames(x))) rownames(x) else 1:nrow(x), 
	id.n, id.method = "y", id.cex = 1, id.col = palette()[1], 
	xlab, ylab, main, what=deparse(substitute(x)), ylim, ...)

Arguments

x

either a numeric data frame or matrix for the default method, or an object of class "mlm" representing a multivariate linear model. In the latter case, residuals from the model are plotted.

...

Other arguments passed to methods

method

estimation method used for center and covariance, one of: "classical" (product-moment), "mcd" (minimum covariance determinant), or "mve" (minimum volume ellipsoid).

detrend

logical; if FALSE, the plot shows values of $D^2$ vs. $\chi^2$. if TRUE, the ordinate shows values of $D^2 - \chi^2$

pch

plot symbol for points

col

color for points; the default is the first entry in the current color palette (see palette and par.

ref.col

Color for the reference line

ref.lwd

Line width for the reference line

conf

confidence coverage for the approximate confidence envelope

env.col

line color for the boundary of the confidence envelope

env.lwd

line width for the confidence envelope

env.lty

line type for the confidence envelope

env.fill

logical; should the confidence envelope be filled?

fill.alpha

transparency value for fill.color

fill.color

color used to fill the confidence envelope

labels

vector of text strings to be used to identify points, defaults to rownames(x) or observation numbers if rownames(x) is NULL

id.n

number of points labeled. If id.n=0, the default, no point identification occurs.

id.method

point identification method. The default id.method="y" will identify the id.n points with the largest value of abs(y-mean(y)). See showLabels for other options.

id.cex

size of text for point labels

id.col

color for point labels

xlab

label for horizontal (theoretical quantiles) axis

ylab

label for vertical (empirical quantiles) axis

main

plot title

what

the name of the object plotted; used in the construction of main when that is not specified.

ylim

limits for vertical axis. If not specified, the range of the confidence envelope is used.

Value

• Returns invisibly the vector of squared Mahalanobis distances corresponding to the rows of x or the residuals of the model.

Details

The calculation of the confidence envelope follows that used in the SAS program, http://www.datavis.ca/sasmac/cqplot.html which comes from Chambers etal. (1983), Section 6.8. The essential formula is $$SE ( z_{(i)} ) = \frac{\hat{\delta}}{g ( q_i )) \times \sqrt{ frac{ p_i (1-p_i} }{n}}$$ where $z_{(i)}$ is the i-th order value of $D^2$, $\hat{\delta}$ is an estimate of the slope of the reference line obtained from the corresponding quartiles and $g(q_i)$ is the density of the chi square distribution at the quantile $q_i$. Note that this confidence envelope applies only to the $D^2$ computed using the classical estimates of location and scatter.

References

J. Chambers, W. S. Cleveland, B. Kleiner, P. A. Tukey (1983). Graphical methods for data analysis, Wadsworth.

See Also

Mahalanobis for calculation of Mahalanobis squared distance qqplot, qqPlot can give a similar result for Mahalanobis squared distances of data or residuals.

Examples

cqplot(iris[, 1:4])

iris.mod <- lm(as.matrix(iris[,1:4]) ~ Species, data=iris)
cqplot(iris.mod, id.n=3)

# compare with car::qqPlot
car::qqPlot(Mahalanobis(iris[, 1:4]), dist="chisq", df=4, id.n=3)