lhist: Histogram with logged x-axis

In FelixTheStudent/ggpower: Package for power transformations, useful e.g. in ggplot

Description

Useful e.g. for log-normal distributions, etc.. Wraps ggplot+geom_histogram with all the things I usually get wrong (see details) Type lhist to see code for copy-pasting.

Usage

lhist(x)

Arguments

x	Numeric vector

Details

Currently, this visualization is the one I find the most useful:

• x-axis is log10-transformed (coord_trans, !not! scale_x - otherwise, adding theoretical probability density distributions does not work)

• log10-spaced bins, otherwise coord_trans throws error and/or bin widths look extremely uneven

• y-axis is sqrt-transformed, otherwise you don't see higher values so well (play around with the example below and you'll see)

• add x-labels with +xlab("var_name") and theoretical distributions with geom_line(aes(logseq(x), dlnorm(x, meanlog, sdlog)))

Value

A ggplot-object.

Examples

## Not run: 
if(interactive()){
 #EXAMPLE1
 x <- rlnorm(100000, meanlog = c(-6,.4), sdlog = 1)
 xseq <- pmax(1e-5, exp(seq(log(.9*min(x)), log(1.1*max(x)), length.out = 100)))
 # with lhist function:
 lhist(x)+
   geom_line(aes(xseq, .5*dlnorm(xseq, -6, 1)), color="blue") +
   geom_line(aes(xseq, .5*dlnorm(xseq, .4, 1)), color="red")
 # without lhist function (slightly more typing):
 ggplot()+
   geom_histogram(data=data.frame(x=x), aes(x, stat(density)), breaks=xseq) +
   geom_line(aes(xseq, .5*dlnorm(xseq, -6, 1)), color="blue") +
   geom_line(aes(xseq, .5*dlnorm(xseq, .4, 1)), color="red") +
   coord_trans(x="log", y="sqrt")
 }

## End(Not run)

FelixTheStudent/ggpower documentation built on March 5, 2020, 8:37 p.m.