rv (version 0.949)

mlplot: Horizontal interval plot of components of a random vector

Description

mlplot plots the scalar components as of the given random array or vector as horizontal intervals. Uncertainty (``credible") intervals are plotted for random components.

Usage

mlplot(X, y.center = TRUE, y.shift = 0, y.map = NULL, mar =
         par("mar"), left.margin = 3, top.axis = TRUE,
         exp.labels = FALSE, x.ticks = NULL, axes = NULL, xlim
         = NULL, ylim = NULL, xlab = deparse(substitute(X)), ylab = NULL,
         las = NULL, add = FALSE, ...)

Arguments

X

a random array or vector

y.center

center the intervals nicely at each y-coordinate?

y.shift

add this amount to each y coordinate of an interval

y.map

optional function to compute the y-coordinates, given X

mar

the margins of the plot

left.margin

offset to add to the left margin of the plot (to add space for the labels)

top.axis

(logical) plot the top axis?

exp.labels

(logical) if the original scale is logarithmic, label ticks in original (exp) scale?

x.ticks

positions for the ticks of the x-axis

axes

(logical) plot the axes at all?

xlim

x limits

ylim

y limits

las

the style of axis labels, see par

add

(logical) add the intervals to an existing plot?

xlab

x label

ylab

not used (instead of labels, the row names are shown)

...

further arguments passed to plot and points

Details

mlplot plots the scalar components of a vector or an array (2 or 3-dimensional) in a vertical grid where each grid cell encloses one component of a vector or a row of a matrix.

Scalars on the same row are plotted tightly together. The positioning of the scalars within a row are controlled by the arguments y.center, y.shift, y.map. These do not need to be set for the default plot; if two arrays or vectors are plotted over on top of each other (using add=TRUE) then you should probably change y.shift which controls the vertical position of the array elements.

See demo(mlplot) for a detailed

To change the color of the random components of the vector, use rvcol. Typically this is of the same length as X, giving the color `theme' for each component.

If X is a 3-dimensional array, mlplot is called repeatedly for each 2-dimensional array X[,,k] for each k.

X may also be a fixed numeric object, but in this case the

NAs (or random scalars with 100% NA) are not plotted.

mlplot is still experimental.

References

Kerman, Jouni and Gelman, Andrew. Manipulating and Summarizing Posterior Simulations Using Random Variable Objects. Technical report, Columbia University, New York.

Examples

n.rows <- 4; n.cols <- 5; n <- (n.rows*n.cols)
  # Draw some fixed numbers
  mu.true <- rnorm(1:n.rows, mean=1:n.rows, sd=1)
  sigma.true <- 1
  theta <- rvmatrix(rvnorm(n=n.cols, mean=mu.true, sd=sigma.true), nrow=n.rows)
  #
  col.labels <- paste("Time", 1:n.cols, sep=":")
  row.labels <- paste("Unit", 1:n.rows, sep=":")
  dimnames(theta) <- list(row.labels, col.labels)
  #
  par(mfrow=c(2,2))
  mlplot(theta, main="theta")
  abline(v=0, lty="dotted")
  mlplot(t(theta), main="theta transposed")
  abline(v=0, lty="dotted")
  row.sd <- apply.rv(theta, 1, sd)
  col.sd <- apply.rv(theta, 2, sd)
  x.max <- max(rvquantile(c(row.sd, col.sd), 0.99))
  mlplot(row.sd, xlim=c(0, x.max), main="theta: within-row sd for each unit")
  abline(v=0)
  mlplot(col.sd, xlim=c(0, x.max), main="theta: between-row sd for each time point")
  abline(v=0)