SMPracticals (version 1.3-1)

pairs.mod: Modified Scatterplot Matrix

Description

Plots a scatterplot matrix in which panels below the diagonal show ordinary scatterplots of pairs of variables, and those above the diagonal show scatterplots of residuals after regression on all the other variables.

Usage

pairs.mod(x, format = "MC", labelnames = names(x), highlight = NULL, level = 0.9, ...)

Arguments

x

A matrix whose rows correspond to units and whose columns correspond to variables measured on those units.

format

'MM' for marginal (that is, standard) scatterplots above and below the diagonal, 'MC' for marginal below and conditional (= partial) above, etc. 'MC' by default.

labelnames

Names of the variables.

highlight

Indexes of observations (rows) to be highlighted.

level

Scalar giving the level for the contour, 0.9 by default.

...

The plotting symbol and other arguments for the points can be controlled by `pch=', etc.

Value

• Produces the scatterplot matrix, and prints the marginal and partial standard deviations of the variables.

Details

The diagonal shows histograms of the original data, and (in black) histograms of the partial residuals after adjustment on all the other variables, shifted to have the same mean as the original data. Also given are the original

The below-diagonal panels contain the numerical value of the correlation, and those above the diagonal contain the partial correlation, that is, the correlation of the residuals after linear regression on the remaining variables. The panels show ellipses which would contain 90 percent of the observations in a large normal sample with the same mean and covariance matrix as the data.

References

Davison, A. C. and Sardy, S. (2000) The partial scatterplot matrix. Journal of Computational and Graphical Statistics, 9, 750--758.

Examples

library(ellipse)
data(mathmarks)
pairs.mod(mathmarks)