[FR] log transformed bins in `histogram`

[fchichorro]

I want to create an histogram for a vector like this:

using UnicodePlots
using Random

Random.seed!(1234)
a = rand(1:5) .^ (0:0.1:5)
histogram(a)

                ┌                                        ┐
   [ 0.0,  5.0)   ┤████████████████████████████████████  24
   [ 5.0, 10.0)  ┤███████████████  10
   [10.0, 15.0) ┤█████████  6
   [15.0, 20.0) ┤██████  4
   [20.0, 25.0) ┤████▌ 3
   [25.0, 30.0) ┤████▌ 3
   [30.0, 35.0) ┤█▌ 1
                └                                        ┘
                                 Frequency

Since the variable is exponential, log transforming it would be more appropriate:

histogram(log10.(a))

              ┌                                        ┐
   [0.0, 0.5) ┤█████████████████████████████████████  8
   [0.5, 1.0) ┤████████████████████████████████▍ 7
   [1.0, 1.5) ┤████████████████████████████████▍ 7
   [1.5, 2.0) ┤████████████████████████████████▍ 7
   [2.0, 2.5) ┤█████████████████████████████████████  8
   [2.5, 3.0) ┤████████████████████████████████▍ 7
   [3.0, 3.5) ┤████████████████████████████████▍ 7
              └                                        ┘
                               Frequency

But I was wondering if there was a way to do it so I do not need to log transform the variable myself. Something that looked like this (notice the bin labels):

              ┌                                        ┐
   [10^0, 10^0.5)    ┤█████████████████████████████████████  8
   [10^0.5, 10^1.0) ┤████████████████████████████████▍ 7
   [10^1.0, 10^1.5) ┤████████████████████████████████▍ 7
   [10^1.5, 10^2.0) ┤████████████████████████████████▍ 7
   [10^2.0, 10^2.5) ┤█████████████████████████████████████  8
   [10^2.5, 10^3.0) ┤████████████████████████████████▍ 7
   [10^3.0, 10^3.5) ┤████████████████████████████████▍ 7
              └                                        ┘
                               Frequency

I tried it with

histogram(a, yscale=:log10)
histogram(a, yscale=x->log10(x))

Interestingly the plot was created as if the yscale argument was not set, which to me was misleading. I think this should either throw an error or the behaviour should be to create logarithmic bins?

This should also probably throw an error:

histogram(a, yscale=:hello)

[t-bltg]

Indeed this would be a great addition.

Horizontal histograms use barplots under the hood, so it's probably best to add an arbitrary scale support there first.

PRs welcome 🙂