plot.CopApprox function

Plot the histogram density approximation to a copula

Plots the histogram density approximation to a copula as determined by GetApprox. Currently works only for bivariate copulae.

## S3 method for class 'CopApprox'
plot(
  x,
  type = c("rgl", "original"),
  col = if (type == "rgl") "#cccccc" else grey.colors(100, start = 0, end = 0.8),
  qcut = 0.95,
  cut,
  alpha = 1,
  topcol = "#ff0000",
  sidecol = "#aaaaaa",
  linecol = "#000000",
  ...
)

Arguments

• x: an object of class CopApprox .
• type: specifies the type of plot to produce. Possible values are ‘rgl’ and ‘original’ ; see Details . Can be abbreviated.
• col: colour(s) to be used; see Details .
• qcut, cut: used if type is ‘rgl’ ; see Details .
• alpha, topcol, sidecol, linecol: used if type is ‘rgl’ ; see Details .
• ...: used if type is ‘original’ ; see Details .

Returns

NULL is returned invisibly .

Details

If type is ‘original’ then plots are produced of the kind shown in Tajvidi and Turlach (2017). In this case the dot arguments are passed to plot when the initial plot is created. Thus, they can be used to set the main title, axes labels and so forth. If the approximation is of type Approximation II, then argument col is used to colour the squares while they are plotted and filled via polygon. For this, the ranks of the probability masses of the squares are mapped (linearly) onto the provided colours.

If type is ‘rgl’ then plots are used using rgl. The code used is based on the ‘hist3d’ demo of the rgl package. The argument col sets the background colour of the plot. Arguments topcol and sidecol are used to set the colour of the top and sides of the cuboids, and the edges of the cuboids are drawn using linecol. Argument alpha sets the transparency. Finally, as the heights of the cuboids can be large, in particular for extreme value copulae, their heights are truncated using either cut (absolute value) or qcut (corresponding quantile of all heights as determined by the quantile function). After truncation, the heights are rescaled to be between 0 and 1, thus the unit on the z -axis is meaningless.

Examples

Cop <- NewMEVGumbelCopula(4)
CopApprox1 <- GetApprox(Cop, dim=2)
plot(CopApprox1)
plot(CopApprox1, type = "o")
CopApprox2 <- GetApprox(Cop, dim=2, type=2)
plot(CopApprox2)
plot(CopApprox2, type = "o", xlab = expression(u[1]), ylab = expression(u[2]))
plot(CopApprox2, type = "o", col = heat.colors(100))

References

Tajvidi, N. and Turlach, B.A. (2017). A general approach to generate random variates for multivariate copulae, Australian & New Zealand Journal of Statistics 60 (1): 140--155. tools:::Rd_expr_doi("10.1111/anzs.12209") .

Author(s)

Berwin A. Turlach berwin.turlach@gmail.com