Re: [Orgmode] org-plot questions

[Eric Schulte]

JBash [EMAIL PROTECTED] writes:

 Hi,

 First of all, thanks to all for the work on org-mode and org-plot.

 A question about org-plot...

 Question 1:
 When I include an index column in the first column of the table,  I am 
 getting evenly spaced data points,
 with the xtic labels being coming from the 'ind:' column (2, in the first 
 data set here).

 The behavior I expected was to use the second column (Frequency) as the 
 independent variable, and to get a
 plot with 3 groups of 3 points, near frequencies of 1, 10, and 20. as I get 
 with the 'Data Set B' group.

 Is this expected and correct?


Hi Jerry,

Not exactly expected, but I believe it is the correct behavior for now.
Notice that you are missing an hline in your table after the headers.
Without an obvious division between the headers, and the table data
org-plot can not tell that the data is numeric, and should be
interpreted as such, so it applies the values of the index column as
xlabels to evenly spaced points (as if they were strings).  Try the
following table, and you should get a result more inline with
expectations.

#+PLOT: ind:2 with:points
| Index | Frequency | Amplitude | Phase | Delta | Peak |
|---+---+---+---+---+--|
| 1 |   0.9 | 9 | 6 | 5 |  198 |
| 2 |   1.0 |10 | 9 | 7 |  212 |
| 3 |   1.1 |11 | 8 | 3 |  144 |
| 4 |   9.9 |19 |12 | 9 |  235 |
| 5 |  10.0 |18 | 7 |12 |  222 |
| 6 |  10.1 |22 |14 | 4 |  187 |
| 7 |  19.9 |14 |11 | 6 |  220 |
| 8 |  20.0 |20 |15 |14 |  213 |
| 9 |  20.1 |18 |16 |11 |  201 |


 I know that this will also include the first column in the plot, which leads 
 to my next question.

 Question 2:
 When I try to use the deps option, I either get an error or an undesired 
 result.  For example, if I include
 a deps:3,4, I get a Wrong type argument error.  If I enclose the 3,4 in 
 single quotes (eg. deps:'3,4'),
 I get only the third column plotted, and not both third and fourth as I 
 expected.  What am I doing wrong?
 A comment - enclosing the argument for deps in single quotes, if that is the 
 intention, seems inconsistent
 with the arguments for other options of that type (ind, for example).


Sorry, the documentation is wrong and it specifies that the dependent
columns should be specified as a comma separated list, when in actuality
they should be specified as a standard lisp list inside parenthesis
separated by spaces.  Thanks for catching this, I will submit a patch
to the documentation.

#+PLOT: ind:1 deps:(3 4)
| Frequency | Amplitude | Phase | Delta | Peak |
|---+---+---+---+--|
|   0.9 | 9 | 6 | 5 |  198 |
|   1.0 |10 | 9 | 7 |  212 |
|   1.1 |11 | 8 | 3 |  144 |
|   9.9 |19 |12 | 9 |  235 |
|  10.0 |18 | 7 |12 |  222 |
|  10.1 |22 |14 | 4 |  187 |
|  19.9 |14 |11 | 6 |  220 |
|  20.0 |20 |15 |14 |  213 |
|  20.1 |18 |16 |11 |  201 |


 Question 3:
 Is there a way to produce 2 different plots from the same table data?


Currently no there is not.  In the current setup all of the plot lines
adjacent to a table are collected and used to generate a single plot for
the table.  I does seem like a good idea to allow multiple plots from a
single table.  Two ideas that come to mind are...

1) keep all of the plot lines adjacent with the table, but number them
   into any number of different plots.  For example the following would
   specify plots 1 and 2 for a table

#+PLOT1: title:the first plot
#+PLOT1: ind:1 with:points
#+PLOT2: title:the second plot on the same table
#+PLOT2: ind:2 deps:(3 4) with:lines
| Frequency | Amplitude | Phase | Delta | Peak |
|---+---+---+---+--|
|   0.9 | 9 | 6 | 5 |  198 |
|   1.0 |10 | 9 | 7 |  212 |
|   1.1 |11 | 8 | 3 |  144 |
|   9.9 |19 |12 | 9 |  235 |
|  10.0 |18 | 7 |12 |  222 |
|  10.1 |22 |14 | 4 |  187 |
|  19.9 |14 |11 | 6 |  220 |
|  20.0 |20 |15 |14 |  213 |
|  20.1 |18 |16 |11 |  201 |

2) the only other mechanism which occurs to me is to somehow label the
   table with a unique ID, and then allow the plot lines to be located
   anywhere in the org-mode file and include the table ID as a new
   argument to the PLOT lines.  This approach seems more flexible if
   more complicated.

Thanks for the bug reports and ideas -- Eric


 Thanks again for the help.
 Jerry

 ** Data Set A
 #+PLOT: ind:2
 | Index | Frequency |  Amplitude | Phase | Delta | Peak |
 | 1 |   0.9 |  9 | 6 | 5 |  

Re: [Orgmode] org-plot questions

[JBash]

Thanks, Eric!

It's a great tool... I like how it, within org, makes it easy to keep the
data organized an yet easily accessible.

Another quick org-plot question...

I found that to turn on the grid, I use:
#+PLOT: set:grid

So, to set an axis label, I tried
#+PLOT: set:xlabel(Frequency)
and
#+PLOT: set:xlabel Frequency
and a few other things, but didn't hit on the right combination.  What's the
correct syntax?

--snip--


 Notice that you are missing an hline in your table after the headers.


Sorry, I should have seen that.


  When I try to use the deps option, I either get an error or an undesired
 result.  For example, if I include
  a deps:3,4, I get a Wrong type argument error.  If I enclose the 3,4
 in single quotes (eg. deps:'3,4'),



 Currently no there is not.  In the current setup all of the plot lines
 adjacent to a table are collected and used to generate a single plot for
 the table.  I does seem like a good idea to allow multiple plots from a
 single table.  Two ideas that come to mind are...

 1) keep all of the plot lines adjacent with the table, but number them
   into any number of different plots.  For example the following would
   specify plots 1 and 2 for a table

 #+PLOT1: title:the first plot
 #+PLOT1: ind:1 with:points
 #+PLOT2: title:the second plot on the same table
 #+PLOT2: ind:2 deps:(3 4) with:lines
 | Frequency | Amplitude | Phase | Delta | Peak |
 |---+---+---+---+--|
 |   0.9 | 9 | 6 | 5 |  198 |
 |   1.0 |10 | 9 | 7 |  212 |
 |   1.1 |11 | 8 | 3 |  144 |
 |   9.9 |19 |12 | 9 |  235 |
 |  10.0 |18 | 7 |12 |  222 |
 |  10.1 |22 |14 | 4 |  187 |
 |  19.9 |14 |11 | 6 |  220 |
 |  20.0 |20 |15 |14 |  213 |
 |  20.1 |18 |16 |11 |  201 |

 2) the only other mechanism which occurs to me is to somehow label the
   table with a unique ID, and then allow the plot lines to be located


This feature (multiple plots from a single table) would be a very nice
addition, with either approach.



 Thanks for the bug reports and ideas -- Eric


You're welcome, and thanks for your help.

Jerry
___
Emacs-orgmode mailing list
Remember: use `Reply All' to send replies to the list.
Emacs-orgmode@gnu.org
http://lists.gnu.org/mailman/listinfo/emacs-orgmode

Re: [Orgmode] org-plot questions

[Samuel Wales]

On Wed, Oct 22, 2008 at 15:43, Eric Schulte [EMAIL PROTECTED] wrote:
 I agree this isn't obvious, if you have a suggestion for improving the
 syntax please let me know.

I haven't looked into this matter, so this comment might not be
relevant, but I do have a suggestion that I had earlier written for
the *general* problem of choosing a syntax for new functionality when
you don't have obvious models to follow.

Some of the desiderata for a syntax include: consistent, easy to
parse, hard to corrupt, simple to quote, simple to escape (especially
with things like regexps that are already complicated), supports being
confident that exporting will not exhibit unexpected behavior, easier
for the user to remember without reference materials, flexible,
extensible, supports macros, nestable, pretty-printable, a published
reference standard, and print-readable (serializing).  If your syntax
has those, then you're doing well.

If you don't have a model to follow (such as org or gnuplot), then
Lisp (el because this is el or cl because it's a standard that is
similar to el and meticulously speced) has usually had a lot of
thought put into its syntax.  You might get a joke or two, but we know
better.  :)

Again, perhaps not relevant, but maybe food for thought for future
questions about syntax.


___
Emacs-orgmode mailing list
Remember: use `Reply All' to send replies to the list.
Emacs-orgmode@gnu.org
http://lists.gnu.org/mailman/listinfo/emacs-orgmode