distribution_mode in somewhat complex cases

[elinw]

If you try some special cases of the mode, I think there are incorrect results.

> distribution_mode(c(1, 2, 3, 3, 4, 4, 5))
[1] 3
> distribution_mode(c(1, 2, 3, 4, 5))
[1] 1
> distribution_mode(c(1, 2, 3, 3, 4, 4, 5, 5))
[1] 3
> distribution_mode(c(1, 2, 3, 3, 4, 5, 5))
[1] 3

On a simple level the issue is what should happen if there are ties or if all of the frequencies are tied. The simplest thing would be to return NA in those cases.

This long Stack Overflow thread
http://stackoverflow.com/questions/2547402/standard-library-function-in-r-for-finding-the-mode

discusses some of the issues.

Update
Here is a link to what SASdoes.
https://documentation.sas.com/doc/en/pgmsascdc/v_073/procstat/procstat_univariate_examples02.htm
Basically, like the current function, when there are multiple modes in a set of statistics they return the lowest but including a note that indicates how many modes there are.
When there are multiple modes and the mode alone is selected they return a table with all of them. (When I made a mode function previously I returned a vector of modes.)