R/lognormal.bin.R

In jrminter/rAnaLab: Helper Functions for the Analytical Laboratory

#' single mode lognormal distribution binning
#'
#'Computes equally-spaced logarithmic bins from a vector
#'and computes the parameters for a single mode lognormal
#'distribution.
#'
#' @param x A vector of observations
#' @param n.root.2 The bin size (nth root of 2, default 8)
#' @param do.plot Boolean (default FALSE) to plot histogram
#' @return A list with a data frame, The mean of log(x), and the stad. dev. of log(x)
#'
#' @examples
#' library(rAnaLab)
#' # data
#' data(diam)
#' l.b <- lognormal.bin(diam[,1])
#' print(str(l.b))
#'
#' @export
#'
#'
lognormal.bin <-
function(x, n.root.2=8.0,  do.plot=FALSE)
{
  x <- subset(x, x>0)
  min.x <- min(x)
  max.x <- max(x)
  # bin by nth root of 2
  log.fact <- log(2.0)/n.root.2

  # print(log.fact)
  l.min.x <- log(min.x)
  l.max.x <- log(max.x)
  n.l.x.max <- ceiling(l.max.x/log.fact)
  n.l.x.min <- floor(l.min.x/log.fact)

  l.bins <- seq( from=n.l.x.min, to=n.l.x.max,
                 by=1) * log.fact
  l.x <- log(x)
  l.mu <- mean(l.x)
  l.sd <- sd(l.x)

  # now compute the histogram
  h <- hist(l.x, breaks=l.bins, plot=do.plot)
  # save what we want
  # midpoints
  h.l.x <- h$mids
  h.x <- exp(h.l.x)
  # counts
  h.cts <- h$counts
  # densities
  h.dens <-h$density
  # make a data frame
  data <- data.frame(log.x=h.l.x,
                     x=h.x,
                     cts=h.cts,
                     dens=h.dens)
  # return a list
  out <-list(data, l.mu, l.sd)
  out
}