Type: Package
Title: A Most Informative Histogram-Like Model
Version: 0.7-1
Date: 2018-05-23
Author: Jouke Witteveen
Maintainer: Jouke Witteveen <j.witteveen@gmail.com>
Description: Using the MDL principle, it is possible to estimate parameters for a histogram-like model. The package contains the implementation of such an estimation method.
Imports: graphics
License: GPL-2 | GPL-3 [expanded from: GPL (≥ 2)]
ByteCompile: yes
NeedsCompilation: yes
Packaged: 2018-05-23 15:11:03 UTC; jouke
Repository: CRAN
Date/Publication: 2018-05-23 15:39:13 UTC

Most Informative Histograms

Description

Compute an estimate of the maximum likelihood parameter of a histogram-like model. If plot=TRUE, the resulting object of class "histogram" is plotted according to plot.histogram, before it is returned.

Usage

histmdl(x, model = "Witteveen", gain = 0, precision = 0, support = 4,
        plot = TRUE, main = paste("Histogram of", xname),
        xlab = xname, ylab = "Density", ...)

Arguments

x

a vector of values for which the histogram is desired.

model

a character string naming the desired histogram-like model. Currently, only "Witteveen" is implemented.

gain

minimum required complexity reduction before an additional interval is accepted.

precision

a value giving the minimum resolution of the data. When computing complexities, boundary values are blurred by this amount to mitigate the effects of, for example, rounding.

support

minimum number of data points desired per interval.

plot

logical. If TRUE (default), a histogram is plotted, otherwise a list of breaks and densities is returned.

main, xlab, ylab

these arguments to title have useful defaults here.

...

further arguments and graphical parameters passed to plot.histogram and thence to title and axis (if plot=TRUE).

Value

an object of class "histogram" which is a list of components:

breaks

the boundaries of intervals. Note that consecutive values are not the same as the boundaries of the intervals that, recursively, define the model instance.

density

densities of the data inside the model-interval that a section is part of.

xname

a character string with the actual x argument name.

Author(s)

Jouke Witteveen and Richard Gill

See Also

hist

Examples

  set.seed (28011988)

  x <- c (rnorm (1000, -6), rnorm (1000, 6))
  histmdl (x, gain=2, col="peru")
  hist (x, freq=FALSE, add=TRUE, col="#80808080")

  x <- c (runif (50), runif (50, max=3))
  histmdl (x, col="peru", ylim=0:1)
  hist (x, freq=FALSE, add=TRUE, col="#80808080")