TkBuildDist: Interactively create a probability distribution.

Description

Build a probability distribution (one option for creating a prior distribution) by clicking or dragging a plot.

Usage

TkBuildDist(x = seq(min + (max - min)/nbin/2, max - (max - min)/nbin/2,
  length.out = nbin), min = 0, max = 10, nbin = 10, logspline = TRUE,
  intervals = FALSE)
TkBuildDist2( min=0, max=1, nbin=10, logspline=TRUE)

Arguments

A starting set of data points, will default to a sequence of uniform values.

min

The minimum value for the histogram

max

The maximum value for the histogram

nbin

The number of bins for the histogram

logspline

Logical, whether to include a logspline curve on the plot and in the output.

intervals

Logical, should the logspline fit be based on the interval counts rather than the clicked data points, also should the interval summary be returned.

Value

Both functions return a list with the breaks that were used the logspline fit (if logspline is TRUE), the x-values clicked on (for TkBuildDist), and the proportion of the distribution within each interval (for TkBuildDist2 or if intervals is TRUE).

Details

Bothe of these functions will open a Tk window to interact with. The window will show a histogram (the defaults will show a uniform distribution), optionally a logspline fit line will be included as well. Including the logspline will slow things down a bit, so you may want to skip it on slow computers.

If you use the TkBuildDist function then a left click on the histogram will add an additional point to the histogram bar clicked on (the actual x-value where clicked will be saved, returned, and used in the optional logspline unless intervals is TRUE). Right clicking on the histogram will remove the point closest to where clicked (based only on x), which will usually have the effect of decreasing the clicked bar by 1, but could affect the neigboring bar if you click near the edge or click on a bar that is 0.

If you use the TkBuildDist2 function then the individual bars can be adjusted by clicking at the top of a bar and dragging up or down, or clicking at what you want the new height of the bar to be. As the current bar is adjusted the other bars will adjust in the oposite direction proportional to their current heights.

The logspline fit assumes the basis for the distribution is the real line, the min and max arguments only control the histogram and where values can be changed.

Examples

Run this code

# NOT RUN {
if(interactive()){
tmp1 <- TkBuildDist()
tmp2 <- TkBuildDist2()
}

# }