crossplot: Bivariate Plot of Posterior Densities

Description

Bivariate plot of the posterior densities of corresponding vectors of parameter nodes. Three plotting methods are provided, that may be overlayed if desired.

If drawcross == TRUE, caterpillar-like plots will be produced, with quantile intervals in the x- and y- directions.
If drawx == TRUE, caterpillar-like plots will be produced, but rotated along the standardized principal component axes. This may be useful to draw if correlation is present.
If drawblob == TRUE, smoothed polygons will be produced, each containing approximately ci= x100% of the associated MCMC samples.

All methods can overlay multiple bars or polygons, depending on the length of ci=.

Usage

crossplot(
  dfx,
  dfy = NULL,
  p = NULL,
  col = 4,
  drawcross = TRUE,
  drawx = FALSE,
  drawblob = FALSE,
  blobres = NULL,
  blobsmooth = NULL,
  outline = FALSE,
  ci = c(0.5, 0.95),
  lwd = 1,
  mean = FALSE,
  link = FALSE,
  linklwd = 1,
  labels = FALSE,
  labelpos = NULL,
  labelcex = 0.7,
  whichx = NULL,
  rowx = NULL,
  columnx = NULL,
  whichy = NULL,
  rowy = NULL,
  columny = NULL,
  xlab = NULL,
  ylab = NULL,
  main = NULL,
  xlim = NULL,
  ylim = NULL,
  transformx = c("none", "exp", "expit"),
  transformy = c("none", "exp", "expit"),
  add = FALSE,
  ...
)

Value

NULL

Arguments

dfx: Output object returned from jagsUI::jags(); or alternately, two-dimensional data.frame or matrix in which parameter node element is given by column and MCMC iteration is given by row. A vector may also be used, that expresses MCMC iterations of a single parameter node. If used with dfy=, this will be plotted in the x-direction.
dfy: Optionally, a two-dimensional data.frame or matrix in which parameter node element is given by column and MCMC iteration is given by row. A vector may also be used, that expresses MCMC iterations of a single parameter node. If used, this will be plotted in the y-direction.
p: Vector of parameter names, if input to dfx is a jagsUI output object. If used, this must be of length 2.
col: Color for plotting, or recyclable vector of colors. Defaults to 4. If col == "random", rcolors will be used to generate a random vector of colors.
drawcross: Whether to draw quantile bars in the x- and y-directions. Defaults to TRUE.
drawx: Whether to draw quantile bars along the standardized principal component axes. Defaults to FALSE.
drawblob: Whether to draw smoothed quantile polygons. Defaults to FALSE.
blobres: Optional tuning parameter for drawing quantile polygons, and corresponds to the number of polygon vertices. If the default NULL is accepted, the function will supply a value based on the number of MCMC samples.
blobsmooth: Optional tuning parameter for drawing quantile polygons, and corresponds to half the number of polygon vertices used for local smoothing. If the default NULL is accepted, the function will supply a value based on the number of MCMC samples and the number of vertices.
outline: Whether to draw quantile polygons as lines rather than filled regions. Defaults to FALSE.
ci: Vector of intervals to overlay. Defaults to 50 percent and 95 percent.
lwd: Base line width for plotting. Defaults to 1.
mean: Whether to include points for means. Defaults to FALSE.
link: Whether to link medians in sequence. Defaults to FALSE.
linklwd: Line width to use for linking. Defaults to 1.
labels: Whether to add labels, or a vector of labels to add. Defaults to FALSE.
labelpos: Optionally, an argument to pos in text for labels. Defaults to NULL.
labelcex: Optional character expansion for labels. Defaults to 0.7.
whichx: Element to subset for x, if only one element of a vector of parameter nodes is desired for plotting.
rowx: Row to subset for x, in the case of a 2-d matrix of parameter nodes in-model.
columnx: Column to subset for x, in the case of a 2-d matrix of parameter nodes in-model.
whichy: Element to subset for x, if only one element of a vector of parameter nodes is desired for plotting.
rowy: Row to subset for y, in the case of a 2-d matrix of parameter nodes in-model.
columny: Column to subset for y, in the case of a 2-d matrix of parameter nodes in-model.
xlab: X-axis label. If the default NULL is accepted, this will be drawn automatically.
ylab: Y-axis label. If the default NULL is accepted, this will be drawn automatically.
main: Plot title.
xlim: X-axis limits. If the default (NULL) is accepted, the limits will be determined automatically.
ylim: Y-axis limits. If the default (NULL) is accepted, the limits will be determined automatically.
transformx: Should the x-axis be (back)transformed? Options are "exp", indicating exponential, or "expit", indicating inverse-logit. Defaults to "none", indicating no transformation. Note: if transformx="exp"is used, consider adding additional plotting argument log="x" or log="xy".
transformy: Should the y-axis be (back)transformed? Options are "exp", indicating exponential, or "expit", indicating inverse-logit. Defaults to "none", indicating no transformation. Note: if transformy="exp"is used, consider adding additional plotting argument log="y" or log="xy".
add: Whether to add to existing plot
...: additional plotting arguments

Author

Matt Tyers

Examples

Run this code

## basic functionality with cross geometry
crossplot(SS_out, p=c("trend","rate"))

## default labels
crossplot(SS_out, p=c("trend","cycle"), labels=TRUE)

## showing:
## - link lines
## - blob geometry (smoothed confidence polygons)
## - random colors with col="random"
crossplot(SS_out, p=c("trend","cycle"),
          labels=SS_data$x, labelpos=1, link=TRUE, drawblob=TRUE,
          col="random")

## adding x geometry and showing usage with a single vector element (41)
crossplot(SS_out, p=c("trend","cycle"),
          whichx=41, whichy=41,
          drawblob=TRUE, drawx=TRUE)

## single vectors (or data.frames or 2d matrices) can also be used
xx <- SS_out$sims.list$trend[,41]
yy <- SS_out$sims.list$cycle[,41]

par(mfrow = c(2, 2))
plot(xx, yy, col=adjustcolor(1, alpha.f=.1), pch=16, main="cross geometry")
crossplot(xx, yy, add=TRUE, col=1)
plot(xx, yy, col=adjustcolor(1, alpha.f=.1), pch=16, main="x geometry")
crossplot(xx, yy, add=TRUE, col=1,
          drawcross=FALSE, drawx=TRUE)
plot(xx, yy, col=adjustcolor(1, alpha.f=.1), pch=16, main="blob geometry")
crossplot(xx, yy, add=TRUE, col=1,
          drawcross=FALSE, drawblob=TRUE)
plot(xx, yy, col=adjustcolor(1, alpha.f=.1), pch=16, main="blob outlines")
crossplot(xx, yy, add=TRUE, col=1,
          drawcross=FALSE, drawblob=TRUE, outline=TRUE)