Re: Candidate stem-and-leaf plot function

[Daniel J Sebald]

On 01/25/2013 10:50 PM, Michael D. Godfrey wrote:
> Attached is an m-file which implements the statistical
> display method called "stem and leaf." This method was
> developed by John Tukey in the mid 1960's. I wrote the
> first machine version (in Fortran, of course). The reference
> for this is John Tukey's book: Exploratory Data Analysis,
> 1971.
>
> There is a stem and leaf plot function in the Matlab Central
> contributed library. I made no attempt at compatibility with this
> implementation. It is not much like what was intended by Tukey.
>
> R also has a routine called stem(). I have not tried this, but
> it is likely well-done. Compatibility with R is not a current Octave
> objective.
>
> I would like comments as to whether this should be included in
> the statistics library, probably in /base for lack of a better place.
> Comments about the code, documentation and tests, are very
> welcome. And, if anyone wants to include it they are welcome to do so.
>
> Michael Godfrey

Interesting.  I'd say that this could be a general plot because it seems 
to me that some applications I've (just now) read about use the plot 
simply to display data, e.g., schedules.  However, I wonder if before 
incorporating into Octave it could be made to create a plot in a figure 
window.  The ASCII-based plot gets the idea across, but if someone wants 
to quickly create a plot that could be incorporated into a document, the 
figure and EPS, PNG, etc. is the way to go.  I'm fairly certain that 
gnuplot has techniques for plotting characters at different locations on 
a graph.  Could color be incorporated in somewhat to enhance meaning?

I had to change "signbit" to "sign" and use 1/-1 in place of 0/1, 
respectively.  Is "signbit" your own routine?  Or am I missing 
something?  That change would be necessary if placed in the base scripts.

One question about the graphing for this example:

x = [-22 12 -28 52  39 -2 12 10 11 11 42 38 44 18 44];
 stem_and_leaf_plot(x, 0);

stem step: 10, data: unsorted.
Hinges:    lo: 10, hi: 42

  -2 | 28
  -1 |
   0 |
   1 | 220118
   2 |
   3 | 98
   4 | 244
   5 | 2

Should the negative stems have the leaves going the other direction?  I.e.,

  -2 | 82

-28 < -22 after all.  I would think for consistency that the quantities 
should always increase when moving outward.  That is, say I added 30 to 
all the values, then the associated leaves would not change positions 
between the two plots.

Dan