inst/shiny/hist-bins/README.md

In jverzani/UsingR: Data Sets, Etc. for the Text "Using R for Introductory Statistics", Second Edition

Example of bin selection when constructing a histogram

Based on example by RStudio.

The basic histogram is defined by:

• a selection of "bins" or sub-intervals of the number to line that encompass the data set

• a count of the observations that land in each bin

• a representation of these counts in terms of bars. If the bins all have the same length, then the height of the bar is traditionally either just the count or may be scaled so the total area of the histogram is $1$.

The choice of bins then is very important. The hist function has several means to specify this:

• the breaks argument can be an algorithm name:

• "Sturges", the default to use $range/(1 + \log(n))$

• "Scott", to use 3.5 * sd(x) * length(x)^(1/3)

• "FD", to use diff(range(x)) / (2 * IQR(x)) * length(x)^(1/3)

The scaling of $n^{1/3}$ is known to be optimal, but the choice of constant makes the latter two different.

As well, the specification of breaks can be a requested number of bins or a specification of the bins.

jverzani/UsingR documentation built on Aug. 3, 2020, 11:57 a.m.