R: Plot Diagnostics for a Bivariate EVD Object

plot.bvevd {evd}

R Documentation

Plot Diagnostics for a Bivariate EVD Object

Description

Six plots (selectable by which) are currently provided: two conditional P-P plots (1,2), conditioning on each margin, a density plot (3), a dependence function plot (4), a quantile curves plot (5) and a spectral density plot (6). Plot diagnostics for the generalized extreme value margins (selectable by mar and which) are also available.

Usage

## S3 method for class 'bvevd'
plot(x, mar = 0, which = 1:6, main, ask = nb.fig <
    length(which) && dev.interactive(), ci = TRUE, cilwd = 1,
    a = 0, grid = 50, legend = TRUE, nplty = 2, blty = 3, method = "cfg",
    convex = FALSE, rev = FALSE, p = seq(0.75, 0.95, 0.05),
    mint = 1, half = FALSE, ...)

Arguments

`x`	An object of class `"bvevd"`.
`mar`	If `mar = 1` or `mar = 2` diagnostics are given for the first or second genereralized extreme value margin respectively.
`which`	A subset of the numbers `1:6` selecting the plots to be shown. By default all are plotted.
`main`	Title of each plot. If given, should be a character vector with the same length as `which`.
`ask`	Logical; if `TRUE`, the user is asked before each plot.
`ci`	Logical; if `TRUE` (the default), plot simulated 95% confidence intervals for the conditional P-P plots.
`cilwd`	Line width for confidence interval lines.
`a`	Passed through to `ppoints` for empirical estimation. Larger values give less probability for extreme events.
`grid`	Argument for the density plot. The (possibly transformed) data is plotted with a contour plot of the bivariate density of the fitted model. The density is evaluated at `grid^2` points.
`legend`	If `legend` is `TRUE` and if the fitted data contained a third column of mode `logical`, then a legend is included in the density and quantile curve plots.
`method`, `convex`, `rev`	Arguments to the dependence function plot. The dependence function for the fitted model is plotted and (optionally) compared to a non-parameteric estimate. See `abvnonpar` for a description of the arguments.
`nplty`, `blty`	Line types for the dependence function plot. `nplty` is the line type of the non-parametric estimate. To omit the non-parametric estimate set `nplty` to zero. `blty` is the line type of the triangular border. To omit the border estimate set `blty` to zero.
`p`, `mint`	Arguments to the quantile curves plot. See `qcbvnonpar` for a description of the plot and the arguments.
`half`	Argument to the spectral density plot. See `hbvevd`.
`...`	Other arguments to be passed through to plotting functions.

Details

In all plots we assume that the fitted model is stationary. For non-stationary models the data are transformed to stationarity. The plot then corresponds to the distribution obtained when all covariates are zero. In particular, the density and quanitle curves plots will not plot the original data for non-stationary models.

A conditional P-P plot is a P-P plot for the condition distribution function of a bivariate evd object. Let G(.|.) be the conditional distribution of the first margin given the second, under the fitted model. Let z_1,\ldots,z_m be the data used in the fitted model, where z_j = (z_{1j}, z_{2j}) for j = 1,\ldots,m. The plot that (by default) is labelled Conditional Plot Two, conditioning on the second margin, consists of the points

\{(p_i, c_i), i = 1,\ldots,m\}

where p_1,\ldots,p_m are plotting points defined by ppoints and c_i is the ith largest value from the sample \{G(z_{j1}|z_{j2}), j = 1,\ldots,m\}. The margins are reversed for Conditional Plot One, so that G(.|.) is the conditional distribution of the second margin given the first.

Examples

bvdata <- rbvevd(100, dep = 0.6, model = "log")
M1 <- fbvevd(bvdata, model = "log")
## Not run: par(mfrow = c(2,2))
## Not run: plot(M1, which = 1:5)
## Not run: plot(M1, mar = 1)
## Not run: plot(M1, mar = 2)