matplotlib-users Mailing List for matplotlib

In the finance .py example, the plots that use collections have integers
plotted on the x-axis.

How do you get dates for x-axis labels?

Bill

zh...@um... wrote:
> How to get the following through? Thanks
> 
> 
> import matplotlib.axes3d as p3
> 
> Traceback (most recent call last):
>   File "<pyshell#19>", line 1, in -toplevel-
>     import matplotlib.axes3d as p3
>   File "C:\Python24\lib\site-packages\matplotlib\axes3d.py", line 26, in
> -toplevel-
>     import art3d
>   File "C:\Python24\lib\site-packages\matplotlib\art3d.py", line 17, in
> -toplevel-
>     import proj3d
>   File "C:\Python24\lib\site-packages\matplotlib\proj3d.py", line 21, in
> -toplevel-
>     cross = nx.cross
> AttributeError: 'module' object has no attribute 'cross'

I don't see that line in the current version, so it looks like you need 
to update your matplotlib installation.

Eric

How to get the following through? Thanks


import matplotlib.axes3d as p3

Traceback (most recent call last):
  File "<pyshell#19>", line 1, in -toplevel-
    import matplotlib.axes3d as p3
  File "C:\Python24\lib\site-packages\matplotlib\axes3d.py", line 26, in
-toplevel-
    import art3d
  File "C:\Python24\lib\site-packages\matplotlib\art3d.py", line 17, in
-toplevel-
    import proj3d
  File "C:\Python24\lib\site-packages\matplotlib\proj3d.py", line 21, in
-toplevel-
    cross = nx.cross
AttributeError: 'module' object has no attribute 'cross'

Second sending (to list instead of gmane) - sorry if this gets through 
twice, but sent it yesterday and until this morning would not see it.  I 
am including indiv. files instead of zip (maybe zip's are blocked?).

Just downloaded the latest build to start testing my things with Python 2.5.

I also installed the examples.zip for 87.1 and found that they were
still using "from wxPython import *" which is deprecated as of wxPython
2.7.  I upgraded some of them to "import wx" and included them in the
attached zip file, there are some (using XRC which I did not upgrade as
I don't know how the XRC stuff works, it also uses the old style import).

Note that the "dynamic_image_wxagg.py" example does not close when
clicking on the "X".  Will keep looking and try to figure out why.

Some of these examples give me the "wxmsg26uh_vc.dll" not found error.
I thought having seen some messages explaining how to deactive the use
of unibol version of wxPython but can't find it.

BTW, will there be a build for Python 2.5 and wxPython 2.7 Unibol (which
is what I used for testing - see below)?

Best regards
Werner

P.S. version info:
# Python 2.5 (r25:51908, Sep 19 2006, 09:52:17) [MSC v.1310 32 bit (Intel)]
# wxPython 2.7.0.1pre.20060923, Boa Constructor 0.4.4

On Friday 29 September 2006 07:50, Jouni K Seppanen wrote:
> Stefan van der Walt <st...@su...> writes:
> > plot(x,y,'o',markerfacecolor='w')
>
> This makes circles filled with white. If you want circles that don't
> obscure whatever is behind them, use markerfacecolor=None.
Jouni, Bill, thank you both for the warning / reminder. I guess Stefan's 
snippet was merely meant as an example to stimulate my memory :-).Everybody 
seeing such an example will start to tinker around and try several things - 
as I did.

Cheers
Christian

Stefan van der Walt <st...@su...> writes:

> plot(x,y,'o',markerfacecolor='w')

This makes circles filled with white. If you want circles that don't
obscure whatever is behind them, use markerfacecolor=None.

-- 
Jouni

Hi,
Is there any way to plot two graphs in two different windows at the same
time? The reason for doing this is that I wanted to update a graph
automatically in a for loop, but it turned out that the first figure has
to closed manually. Any suggestion for that is also greatly appreciated.
Best Regards, John

>>>>> "Christian" == Christian Meesters <mee...@un...> writes:

    Christian> One last remark on this: Since so many journals demand
    Christian> this, would it be worth a feature request? (I don't
    Christian> have the time nor the skills to work on this.)  One may
    Christian> think of this as one wishes (don't like these
    Christian> publishing companies and their demands, either), but
    Christian> matplotlib wants to enable its users to produce figures
    Christian> ready for publication, right?

Certainly -- just add a feature request to the sf site and post the
same to the users or devel list.  Can't promise that we have the time
either (though we may have the skills) but maybe someone will take it
up.  Darren Dale has done the most work of late on the PS backend and
he is currently on vacation -- maybe when he returns he'll take it up.

JDH

On Thursday 28 September 2006 1:24 pm, Christian Meesters wrote:
> Hi,
>
> I'd like to plot experimental data points with fitted data through
> it. This time best would be to plot hollow circles for the
> experimental data. Pretty much like literal 'o's (except, of
> course, that passing 'o' results in thick circles).
> Is this possible somehow?

Hi Christian,


I've been using the Rectangle class in patches to plot unfilled 
rectangles. Circle and Polygon classes are also available that you 
may find useful.

You can control the filling, the colors of the faces and edges, and 
the thickness of the edges.

Here's an example for an unfilled red rectangle.


from matplotlib.patches import Rectangle
...
...
fig=figure(figsize=(W,H))
ax = fig.add_axes([left,bottom,width,height])        
p=Rectangle([x,y], 4.0, 0.6, fill=False, linewidth=1, edgecolor="red")
ax.add_patch(p)


-Cheers

FYI I don't know what the current state of affairs was but I
looked into this a little bit about 5 or so years ago.  Adobe has
a spec for what the preview image in an EPS file is suppose to be
(I can't remember the details on what it is but you can google
for the EPS spec) but I don't think anyone really followed it. 
On Mac's, everything produced a PICT image and on Windows
everything produced something else (I forget exactly what it
was).  In short, at the time I checked it out, it was very
difficult to make a high quality preview image that would work
well on a Mac and a Windows box. 

It would be nice if things were different now.  We do a lot of
plotting that ends up in presentations and it would be nice if
you could embed an EPS in the presentation with a high quality
preview.

Ted

---------Included Message----------
>Date: 28-sep-2006 14:04:55 -0700
>From: "Christian Meesters" <mee...@un...>
>To: <mat...@li...>
>Subject: Re: [Matplotlib-users] eps file format
>
>> As for the preview header, I suspect there arte 3rd part tools
that
>> can do this (ImageMagick?).  We should be able to do it
ourself with
>> agg, but it would require someone to dig in and figure out the
spec.
>>
>> JDH
>One last remark on this: Since so many journals demand this,
would it be worth 
>a feature request? (I don't have the time nor the skills to work
on this.) 
>One may think of this as one wishes (don't like these publishing
companies 
>and their demands, either), but matplotlib wants to enable its
users to 
>produce figures ready for publication, right?
>
>Christian 
>
>-------------------------------------------------------------------------
>Take Surveys. Earn Cash. Influence the Future of IT
>Join SourceForge.net's Techsay panel and you'll get the chance
to share your
>opinions on IT & business topics through brief surveys -- and
earn cash
>http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
>_______________________________________________
>Matplotlib-users mailing list
>Mat...@li...
>https://lists.sourceforge.net/lists/listinfo/matplotlib-users
>
>
---------End of Included Message----------

>
> I'm sure someone will soon provide you with an insightful answer.  In
> the meantime, you can fudge it by doing
>
> plot(x,y,'o',markerfacecolor='w')

And this is not an "insightful answer"? Anyway, thanks a lot!
Christian (who was to tired to look at the right place ...)

On Thu, Sep 28, 2006 at 08:24:44PM +0200, Christian Meesters wrote:
> I'd like to plot experimental data points with fitted data through it. =
This=20
> time best would be to plot hollow circles for the experimental data. Pr=
etty=20
> much like literal 'o's (except, of course, that passing 'o' results in =
thick=20
> circles).
> Is this possible somehow?

I'm sure someone will soon provide you with an insightful answer.  In
the meantime, you can fudge it by doing

plot(x,y,'o',markerfacecolor=3D'w')
plot(x,y)

Regards
St=E9fan

> As for the preview header, I suspect there arte 3rd part tools that
> can do this (ImageMagick?).  We should be able to do it ourself with
> agg, but it would require someone to dig in and figure out the spec.
>
> JDH
One last remark on this: Since so many journals demand this, would it be worth 
a feature request? (I don't have the time nor the skills to work on this.) 
One may think of this as one wishes (don't like these publishing companies 
and their demands, either), but matplotlib wants to enable its users to 
produce figures ready for publication, right?

Christian 

Hi,

I'd like to plot experimental data points with fitted data through it. This 
time best would be to plot hollow circles for the experimental data. Pretty 
much like literal 'o's (except, of course, that passing 'o' results in thick 
circles).
Is this possible somehow?

TIA
Christian

Thanks, John and Alan,

That was fast!

> As for the fonts, yes, you can control this with rc.
Sure, but what about the latex rendered parts? (I wonder whether it actually 
matters, but I'd like to be sure, because there so little time left ...)

> On Thu, 28 Sep 2006, Christian Meesters apparently wrote:
> > - it should include a 8bit preview/header at a resolution of 72dpi - no
> > idea how to generate this
>
> I think you can use
> http://www.cs.wisc.edu/~ghost/gsview/epstool.htm
> but do not know what dpi control you'll have.
It works, thanks. The installation is easy. Setting the resolution in dpi can 
be done by giving the flag --dpi <resolution>. 

Cheers
Christian

On Thu, 28 Sep 2006, Christian Meesters apparently wrote: 
> - it should include a 8bit preview/header at a resolution of 72dpi - no idea 
> how to generate this 

I think you can use
http://www.cs.wisc.edu/~ghost/gsview/epstool.htm
but do not know what dpi control you'll have.

hth,
Alan Isaac

>>>>> "Christian" == Christian Meesters <mee...@un...> writes:

    Christian> Hi, I'd like to write a paper with figures in eps
    Christian> format. Since the paper is for one of Elseviers
    Christian> journals the eps-figures have to meet the following
    Christian> requirements: - all colors have to be in the rgb color
    Christian> space - this seems to be true for the files I generate.
    Christian> - it should include a 8bit preview/header at a
    Christian> resolution of 72dpi - no idea how to generate this -
    Christian> allowed fonts are: Arial, Courier, Helvetica, Times,
    Christian> Symbol - this is something I can set in the rc file,
    Christian> although I have no idea what is true in latex rendered
    Christian> parts

    Christian> Can somebody tell me please, whether matplotlib indeed
    Christian> creates colors in the rgb color space and answer the
    Christian> other two questions, too? Would be really great ...

Yes, it does use the RGB colorspace, but we would like to support
other colorspaces too.

As for the fonts, yes, you can control this with rc.

As for the preview header, I suspect there arte 3rd part tools that
can do this (ImageMagick?).  We should be able to do it ourself with
agg, but it would require someone to dig in and figure out the spec.

JDH

Hi,

I'd like to write a paper with figures in eps format. Since the paper is for 
one of Elseviers journals the eps-figures have to meet the following 
requirements:
- all colors have to be in the rgb color space - this seems to be true for the 
files I generate.
- it should include a 8bit preview/header at a resolution of 72dpi - no idea 
how to generate this
- allowed fonts are: Arial, Courier, Helvetica, Times, Symbol - this is 
something I can set in the rc file, although I have no idea what is true in 
latex rendered parts

Can somebody tell me please, whether matplotlib indeed creates colors in the 
rgb color space and answer the other two questions, too? Would be really 
great ...

TIA
Christian

Attached is an updated version which does close.

I move the timer code and added and EVT_CLOSE which stops the timer and 
now it closes correctly.

Werner

Hi, All,

 

I am doing some plotting and I want to do some format change to my tick.
Here is what I want:

 

Origin ticks:      0.0             1.0             2.0             3.0
4.0             5.0

Ticks I want:      0.0mV       1000mv    2000mV   3000mV   4000mV   5000mV

 

Just showing the ticks in an easier-read way. Because I don't want so see
anything like 1.2345e-2, I want to see 12.345mV.

 

Thanks very much..

 

CONFIDENTIALITY NOTE: This message is intended only for the use of the 
individual or entity to which it is addressed and may contain information 
that is privileged, confidential, and exempt from disclosure under 
applicable law. If the reader of this notice is not the intended recipient, 
you are hereby notified that any dissemination, distribution or copying of 
this communication is strictly prohibited. If you have received this 
communication in error, please notify the sender immediately by telephone at

(86-10)8225 5779-361.

 



_____________________________________________________________________________
Scanned by IBM Email Security Management Services powered by MessageLabs. For more information please visit http://www.ers.ibm.com
_____________________________________________________________________________

John,

Thanks for the information.  I might start looking at the code and 
see if I can make any changes.  If I'm successful then I'll submit 
these as patches.

Gerardo


>
>
>No, there isn't, but patches here would be most welcome since this has
>come up before....
>
>JDH

>>>>> "Gerardo" == Gerardo Rivera <Ger...@jp...> writes:

    Gerardo> Hi, I'm currently using matplotlib with Plone/Zope to
    Gerardo> generate some graphics for temperature, wind speed and
    Gerardo> wind direction based on some data from Lake Tahoe.
    Gerardo> However, the currently layout of the legends box really
    Gerardo> restricts me from displaying the full set of data points
    Gerardo> on one graph.

    Gerardo> Currently the legend layout is displayed as:

    Gerardo> line1_color label1 line2_color label2 line3_color label3
    Gerardo> ...

    Gerardo> we want horizontal layout:

    Gerardo> line1_color label1 line2_color label2 line3_color label3
    Gerardo> ...

    Gerardo> Also placing the horizontal layout below the x-axis title
    Gerardo> would be ideal.

    Gerardo> Is there a way to create the horizontal layout in the
    Gerardo> current version of matplotlib?

No, there isn't, but patches here would be most welcome since this has
come up before....

JDH

Hi,

I'm currently using matplotlib with Plone/Zope to generate some 
graphics for temperature, wind speed and wind direction based on some 
data from Lake Tahoe.  However, the currently layout of the legends 
box really restricts me from displaying the full set of data points 
on one graph.

Currently the legend layout is displayed as:

line1_color   label1
line2_color   label2
line3_color   label3
...

we want horizontal layout:

line1_color   label1        line2_color   label2        line3_color 
label3 ...

Also placing the horizontal layout below the x-axis title would be ideal.

Is there a way to create the horizontal layout in the current version 
of matplotlib?

I noticed that someone created a column layout in Aug 2006 but 
actually changed the legend code.  If I need to I could look at the 
legend code myself to get the needed results.

I'm using the following:

matplotlib 0.87.5
python 2.3.5, win32
Zope 2.7.8
Plone 2

Thanks for any help
Gerardo Rivera

>>>>> "Scott" == Scott Ransom <sr...@nr...> writes:

    Scott> Hi All, I think I might have uncovered a bug in the legend
    Scott> code when using multiple patches so that only the first
    Scott> patch type is used in the legend.

    Scott> In [41]: matplotlib.__version__ Out[41]: '0.87.5'

    Scott> (that is revision 2782 from SVN)

    Scott> Here is some code that shows the problem:

    Scott> ------------------------------ from pylab import *

    Scott> binctrs = linspace(-4.0, 4.0, 15) vals = randn(1000) n1,
    Scott> bins1, p1 = hist(vals, binctrs, align='center') n2, bins2,
    Scott> p2 = hist(vals[:300], binctrs, align='center') setp(p1,
    Scott> 'facecolor', 'k', 'alpha', 0.2) setp(p2, 'facecolor', 'g',
    Scott> 'alpha', 0.9) legend((p1, p2), ("Frobs", "NewFrobs"))
    Scott> show() ------------------------------

The returned patches is a list of patches -- you just need to pick off
a representative one

ax = subplot(111)
binctrs = linspace(-4.0, 4.0, 15)
vals = randn(1000)
n1, bins1, p1 = hist(vals, binctrs, align='center', facecolor='black', alpha=0.2)
n2, bins2, p2 = hist(vals[:300], binctrs, align='center', facecolor='green', alpha=0.9)
legend((p1[0], p2[0]), ("Frobs", "NewFrobs"))
show()

Please disregard, i finally solved my problem using IndexLocator:

# datetime variable x is in 15 min intervals, 6 1/2 hours in trading day
# results in 26 data points per day, so:
days = IndexLocator(26, 1)
hours = IndexLocator(4, 1)
ax = subplot(111)
plot(datesindex, y)

ax.xaxis.set_major_locator(days)
ax.xaxis.set_major_formatter(IndexDateFormatter(x, '%Y-%m-%d %H:%M:%S'))
ax.xaxis.set_minor_locator(hours)
ax.xaxis.set_minor_formatter(IndexDateFormatter(x, 'Y-%m-%d %H:%M:%S'))

Thanks to all who have help me with previous issues. :)

On Wed, 2006-09-27 at 09:58 -0400, Richard Albright wrote:
> I am getting closer to the results I want by modifying my days and hours
> locators to read:
> 
> days = WeekdayLocator(byweekday=(MO, TU, WE, TH, FR))
> hours = HourLocator(byhour=range(9,17,1))
> 
> but the xaxis still fills in times for the rest of each day interval,
> but now only shows xtick labels for the period during the day i am
> concerned with.  Is there a way to limit the hours plotted on the xaxis
> for each day interval?
> 
> 
> On Tue, 2006-09-26 at 16:47 -0400, Richard Albright wrote:
> > I am trying to figure out how i can use IndexLocator with DateFormatter.
> > Currently i am only using DayLocator and HourLocator in my code, but it
> > is showing ticks for all hours in the plot, when I only care about
> > plotting the hours for when data is available.
> > 
> > I already applied date2num to my dates in the x list below.
> > 
> > 
> > 
> > from matplotlib.dates import IndexDateFormatter, DateFormatter, date2num
> > 
> > import matplotlib
> > 
> > from matplotlib import rc
> > 
> > import datetime
> > from pylab import *
> > 
> > 
> > x = [732559.39583333337, 732559.40625, 732559.41666666663,
> > 732559.42708333337,
> >     732559.4375, 732559.44791666663, 732559.45833333337, 732559.46875, 
> >     732559.47916666663 , 732559.48958333337, 732559.5,
> > 732559.51041666663, 
> >     732559.52083333337, 732559.53125, 732559.54166666663,
> > 732559.55208333337,
> >     732559.5625, 732559.57291666663, 732559.58333333337, 732559.59375, 
> >     732559.60416666663, 732559.61458333337, 732559.625,
> > 732559.63541666663, 
> >     732559.64583333337, 732559.65625, 732559.66666666663,
> > 732560.39583333337, 
> >     732560.40625, 732560.41666666663, 732560.42708333337, 732560.4375, 
> >     732560.44791666663, 732560.45833333337, 732560.46875,
> > 732560.47916666663, 
> >     732560.48958333337, 732560.5, 732560.51041666663,
> > 732560.52083333337, 
> >     732560.53125, 732560.54166666663, 732560.55208333337, 732560.5625, 
> >     732560.57291666663, 732560.58333333337, 732560.59375,
> > 732560.60416666663, 
> >     732560.61458333337, 732560.625, 732560.63541666663,
> > 732560.64583333337, 
> >     732560.65625, 732560.66666666663]
> > y = [13.07, 13.050000000000001, 13.029999999999999, 13.039999999999999, 
> >     13.029999999999999, 13.0, 13.0, 13.0, 13.0, 13.029999999999999,
> > 13.02, 
> >     13.01, 12.99, 12.98, 12.98, 12.960000000000001, 12.94, 12.92, 
> >     12.960000000000001, 12.98, 13.19, 13.140000000000001,
> > 13.130000000000001, 
> >     13.109999999999999, 13.08, 13.109999999999999, 13.039999999999999,
> > 13.15, 
> >     13.130000000000001, 13.15, 13.279999999999999, 13.210000000000001, 
> >     13.199999999999999, 13.199999999999999, 13.19, 13.199999999999999, 
> >     13.140000000000001, 13.119999999999999, 13.1, 13.09,
> > 13.130000000000001, 
> >     13.119999999999999, 13.130000000000001, 13.130000000000001, 
> >     13.140000000000001, 13.15, 13.16, 13.119999999999999,
> > 13.119999999999999, 
> >     13.1, 13.050000000000001, 13.029999999999999, 13.02, 13.0]
> > 
> > datesindex = range(len(x))
> > 
> > days = DayLocator()
> > hours = HourLocator()
> > ax = subplot(111)
> > plot(x, y)
> > 
> > ax.xaxis.set_major_locator(days)
> > ax.xaxis.set_major_formatter(DateFormatter('%Y-%m-%d %H:%M:%S'))
> > ax.xaxis.set_minor_locator(hours)
> > ax.xaxis.set_minor_formatter(DateFormatter('%H:%M:%S'))
> > setp(ax.get_xticklabels(), 'rotation', 90, 'horizontalalignment',
> > 'center', fontsize=8)
> > minorlabels = [tick.label1 for tick in ax.xaxis.get_minor_ticks()]
> > setp(minorlabels, 'rotation', 90, 'horizontalalignment', 'center',
> > fontsize=8)
> > 
> > grid(True)
> > show()
> > 
> > Any suggestions would be greatly appreciated.
> > 
> -- 
> Rick Albright
> Senior Quantitvative Analyst
> Indie Research, LLC
> 254 Witherspoon Street
> Princeton, NJ 08542
> (609)497-1030
> ral...@in...
> 
> 
> 
> -------------------------------------------------------------------------
> Take Surveys. Earn Cash. Influence the Future of IT
> Join SourceForge.net's Techsay panel and you'll get the chance to share your
> opinions on IT & business topics through brief surveys -- and earn cash
> http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
> _______________________________________________
> Matplotlib-users mailing list
> Mat...@li...
> https://lists.sourceforge.net/lists/listinfo/matplotlib-users
-- 
Rick Albright
Senior Quantitvative Analyst
Indie Research, LLC
254 Witherspoon Street
Princeton, NJ 08542
(609)497-1030
ral...@in...