Andrews' Curves

Plots Andrews' curves in cartesian or polar coordinates.

andrewsplot(A, f, style = "pol", scaled = FALSE, npts = 101)

Arguments

A

numeric matrix with at least two columns.

f

factor or integer vector with nrow(A) elements.

style

character variable, only possible values `cart' or `pol'.

scaled

logical; if true scales each column to have mean 0 and standard deviation 1 (not yet implemented).

npts

number of points to plot.

Details

andrewsplot creates an Andrews plot of the multivariate data in the matrix A, assigning different colors according to the factor or integer vector f.

Andrews' plot represent each observation (row) by a periodic function over the interval [0, 2*pi]. This function for the i-th observation is defined as ...

The plot can be seen in cartesian or polar coordinates — the latter seems appropriate as all these functions are periodic.

Value

Generates a plot, no return value.

Note

Please note that a different ordering of the columns will result in quite different functions and overall picture.

There are variants utilizing principal component scores, in order of decreasing eigenvalues.

References

R. Khattree and D. N. Naik (2002). Andrews PLots for Multivariate Data: Some New Suggestions and Applications. Journal of Statistical Planning and Inference, Vol. 100, No. 2, pp. 411–425.

See also

polar, andrews::andrews

Examples

if (FALSE) { # \dontrun{
data(iris)
s <- sample(1:4, 4)
A <- as.matrix(iris[, s])
f <- as.integer(iris[, 5])
andrewsplot(A, f, style = "pol")
} # }