possible glitches [was: Patch for new "stats" command]

[Ethan Merritt]

 On Sunday 08 November 2009, Philipp K. Janert wrote:
> We'd like to hear feedback and suggestions. Is this
> useful? Are we missing anything?
 
Some first thoughts:
=======================
-  int_error( NO_CARET,
              "Out of memory in stats: too many datapoints (%d)?", max_n );
   Can this really happen?   Does the session survive?


=========================
- What's the intent with binary data?  Simple attempt based on image.dem
  doesn't work:

gnuplot> plot 'blutux.rgb' binary array=(128,128) dx=1 flip=y rotation=0.5pi \
               origin=(0,0) format='%uchar' using ($1+$2+$3)/3 with image
gnuplot> stats 'blutux.rgb' binary array=(128,128) format='%uchar' using ($1+$2+$3)/3
                                                                                                                              ^
         Need 0 to 2 using specs for stats command


========================
- I don't know what values are being pulled from this data,
  but the stats don't look right :-)

gnuplot> stats '-' matrix
input data ('e' ends) > 1 2 3 4
input data ('e' ends) > 1 2 3 4
input data ('e' ends) > 1 2 3 4
input data ('e' ends) > 1 2 3 4
input data ('e' ends) > e      
input data ('e' ends) > e      

* FILE:
  Records:      16
  Invalid:       0
  Blank:         3
  Data Blocks:   1

* COLUMNS:
  Mean:          1.5000             2.5000
  Std Dev:       1.1180             1.1180
  Sum x:        24.0000            40.0000
  Sum x^2:      56.0000           120.0000

  Minimum:       0.0000 [ 1]        1.0000 [ 1]
  Quartile:      1.0000 [ 2]        2.0000 [ 2]
  Median:        2.0000 [ 3]        3.0000 [ 3]
  Quartile:      3.0000 [ 4]        4.0000 [ 4]
  Maximum:       3.0000 [16]        4.0000 [16]




------------------------------------------------------------------------------
Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day
trial. Simplify your report design, integration and deployment - and focus on
what you do best, core application coding. Discover what's new with
Crystal Reports now.  http://p.sf.net/sfu/bobj-july
_______________________________________________
gnuplot-beta mailing list
gnuplot-beta@...
https://lists.sourceforge.net/lists/listinfo/gnuplot-beta

[Philipp K. Janert]

On Sunday 08 November 2009 09:14:36 pm Ethan Merritt wrote:
>  On Sunday 08 November 2009, Philipp K. Janert wrote:
> > We'd like to hear feedback and suggestions. Is this
> > useful? Are we missing anything?
>
> Some first thoughts:
> =======================
> -  int_error( NO_CARET,
>               "Out of memory in stats: too many datapoints (%d)?", max_n );
>    Can this really happen?   Does the session survive?

Probably not. But then again, if you are out of memory
here, there is probably little useful you can do with the
session going forward. For me, the purpose of this
command was to at least tell the user what was wrong,
rather than dying silently.

Regarding binary and matrix data: If you look at the
documentation, you will notice that we dont' claim
to support either binary nor matrix data. I guess
we should have been more explicit about flagging
those keywords as errors on the command line.

That is a limitation, but it's not clear to me how much
of a real limitation this is in practice. How important
is the matrix format overall, and in particular outside
of splot?

(I think the "fit" command does not handle either
format, either.)

>
>
> =========================
> - What's the intent with binary data?  Simple attempt based on image.dem
>   doesn't work:
>
> gnuplot> plot 'blutux.rgb' binary array=(128,128) dx=1 flip=y
> rotation=0.5pi \ origin=(0,0) format='%uchar' using ($1+$2+$3)/3 with image
> gnuplot> stats 'blutux.rgb' binary array=(128,128) format='%uchar' using
> ($1+$2+$3)/3 ^ Need 0 to 2 using specs for stats command
>
>
> ========================
> - I don't know what values are being pulled from this data,
>   but the stats don't look right :-)
>
> gnuplot> stats '-' matrix
> input data ('e' ends) > 1 2 3 4
> input data ('e' ends) > 1 2 3 4
> input data ('e' ends) > 1 2 3 4
> input data ('e' ends) > 1 2 3 4
> input data ('e' ends) > e
> input data ('e' ends) > e
>
> * FILE:
>   Records:      16
>   Invalid:       0
>   Blank:         3
>   Data Blocks:   1
>
> * COLUMNS:
>   Mean:          1.5000             2.5000
>   Std Dev:       1.1180             1.1180
>   Sum x:        24.0000            40.0000
>   Sum x^2:      56.0000           120.0000
>
>   Minimum:       0.0000 [ 1]        1.0000 [ 1]
>   Quartile:      1.0000 [ 2]        2.0000 [ 2]
>   Median:        2.0000 [ 3]        3.0000 [ 3]
>   Quartile:      3.0000 [ 4]        4.0000 [ 4]
>   Maximum:       3.0000 [16]        4.0000 [16]

------------------------------------------------------------------------------
Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day
trial. Simplify your report design, integration and deployment - and focus on
what you do best, core application coding. Discover what's new with
Crystal Reports now.  http://p.sf.net/sfu/bobj-july
_______________________________________________
gnuplot-beta mailing list
gnuplot-beta@...
https://lists.sourceforge.net/lists/listinfo/gnuplot-beta

[Ethan Merritt]

On Sunday 08 November 2009, Philipp K. Janert wrote:
>
> Regarding binary and matrix data: If you look at the
> documentation, you will notice that we dont' claim
> to support either binary nor matrix data. I guess
> we should have been more explicit about flagging
> those keywords as errors on the command line.

Binary data is important, I think.
You should get binary support for free since you are calling
df_readline(), which already knows how to deal with binary input.

I think the glitch is not because it's binary, but because it's
an array of data values.  For the purpose of finding min/max/mean
we don't care about that, but the program returns the values
internally via column 3 rather than column 1.  
I think... I get confused every time I have to use a "using" statement
with matrix or array input data.


> How important is the matrix format overall, and in particular
> outside of splot?

Doesn't have anything to do with splot.

Think about setting the color palette for a 2D heat map based
on the stats of the data being displayed.  As the program is
now, you can set the cbrange manually based on prior knowledge
or you can autoscale to the full min/max range.  It would be
very much in the spirit of your "stats" patch to facilitate
setting the cbrange based on mean/stddev.

        Ethan

------------------------------------------------------------------------------
Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day
trial. Simplify your report design, integration and deployment - and focus on
what you do best, core application coding. Discover what's new with
Crystal Reports now.  http://p.sf.net/sfu/bobj-july
_______________________________________________
gnuplot-beta mailing list
gnuplot-beta@...
https://lists.sourceforge.net/lists/listinfo/gnuplot-beta