I recently gave a talk about Data Analysis on the command line at
Bitreich Conference 2019 [1]. I know it might seem strange, but the
shell (and the mechanisms it provides, and especially pipes) is indeed
well-suited for a variety of data analysis tasks. Most of the times, you
can just get going with standard Unix tools and produce some elaborated
results by just piping one command after the other.

One important part of data analysis is the presentation of your results.
`gnuplot(1)` is a very nice tool to produce plots. And it can also
easily generate nice ASCII plots. 

In the talk I showed how to generate a simple ASCII histogram out of a
data set reporting the number of chess games played by players in each
range of Elo rating. Forget the details of how the data set was obtained
(or have a look at the slides of the talk for more info), and just focus
on the data, which in the end is in the format:

   404184 20
   548579 21
   670332 22
   714168 23
   717261 24
   534279 25
   246087 26
   75139 27
   6114 28
 
This means that 404184 games were played by players with an Elo rating
between 2000 and 2099. Then 548579 games were player by players with an
Elo rating between 2100 and 2199, and so on. In `gnuplot` you can plot
an ASCII histogram of this data set using:

  $ gnuplot
  > set term dumb size 80,20
  > set xrange [19:29]
  > plot "data.txt" u 2:1 w impulses title "player rating"

  800000 +------------------------------------------------------------------+   
         |      +            +             +            +            +      |   
  700000 |-+                        *      *          player rating *******-|   
         |                   *      *      *                                |   
  600000 |-+                 *      *      *                              +-|   
         |            *      *      *      *     *                          |   
  500000 |-+          *      *      *      *     *                        +-|   
         |            *      *      *      *     *                          |   
  400000 |-+    *     *      *      *      *     *                        +-|   
         |      *     *      *      *      *     *                          |   
  300000 |-+    *     *      *      *      *     *                        +-|   
         |      *     *      *      *      *     *      *                   |   
  200000 |-+    *     *      *      *      *     *      *                 +-|   
         |      *     *      *      *      *     *      *                   |   
  100000 |-+    *     *      *      *      *     *      *      *          +-|   
         |      *     *      *      *      *     *      *      *     +      |   
       0 +------------------------------------------------------------------+   
                20           22            24           26           28         

As usual, more options and customisations are available, if you RTFM.

  -+-+-+-

gnuplot(1) was initially developed by Thomas Williams and Colin Kelley
           in 1986 [2]. Despite its name, it has nothing to do with the
	   GNU project, and is distributed under separate a non-copyleft 
	   free software licence (the Gnuplot License [2]).

  -+-+-+-

[1] gopher://bitreich.org/0/con/2019/rec/data-analysis.md

[2] http://www.gnuplot.info/

[3] https://directory.fsf.org/wiki/License:Gnuplot