matplotlib-users Mailing List for matplotlib

hi all,

can someone advise on how to make simple venn diagrams, like the one here:

http://en.wikipedia.org/wiki/File:Venn_diagram_cmyk.svg

simply three (or more) intersecting circles, such that one can label every
point of their intersection, and maybe make the circles in size proportion
to the number of elements they are supposed to represent.  i know some
people use Sage for this but i prefer to use matplotlib directly.

any help / info on how to get started on this or some example code would be
greatly appreciated.

thank you.

per freem wrote:
> hi all,
> 
> please disregard the previous email - i had a mistake in my file that 
> did not do the casting properly when loading the data.
> 
> i managed to plot my data, but this time i am having a problem with the 
> 'bar' function.
> 
> when i plot using:
> 
> x = data[:, 0]
> y = data[:, 1]
> bar(x,y)
> 
> i get the attached figure. the bar graphs are way too thin and don't 
> look like bar graphs at all. i see in the gallery many examples of bars 
> with greater width, e.g. 
> http://matplotlib.sourceforge.net/examples/api/histogram_demo.html
> 
> but all of these seem to be made using the 'hist' function. i just want 
> the bar width to be greater. my setting of the width= does not make a 
> difference, it treats:
> 
> bar(x,y,width=1.5)
> bar(x,y,width=10)
> etc.

Width is in the same units as x, and it looks like your range of x 
values is 1e8, so maybe your width needs to be something like 1e7, not 10.

Eric
> 
> as the same, yielding this line plot. if i remove some data points (and 
> plot x and y's that are only, say, 3 in length) then the bars look normal.
> 
> how can i make the bar widths greater in this case?
> 
> On Sun, Mar 1, 2009 at 11:41 AM, per freem <per...@gm... 
> <mailto:per...@gm...>> wrote:
> 
>     hi all,
> 
>     i am reading a set of tab-separated data from a file and i want to
>     put it into an array, and then plot some of the columns. i know the
>     number of columns ahead of time but not the number of rows. i load
>     the array from the file as follows, which seems to work:
> 
>     data = []
>     for line in myfile:
>       field1, field2, field3 = line.strip().split('\t')
>       data.append([int(field1), int(field2), int(field3)])
> 
>     i then convert it into an array as follows:
> 
>     data = array(data)
> 
>     i am able to reference the first column as follows:
> 
>     data[:,0]
> 
>     but if i try to plot the first column against the second as follows:
> 
>     bar(data[:,0],data[:,1])
> 
>     then i get the error:
> 
>     /usr/lib64/python2.5/site-packages/matplotlib/units.pyc in
>     get_converter(self, x)
>         128             converter = self.get(classx)
>         129
>     --> 130         if converter is None and iterable(x):
>         131             # if this is anything but an object array, we'll
>     assume
>         132             # there are no custom units
> 
>     [repeated many times]
> 
>     RuntimeError: maximum recursion depth exceeded
>     WARNING: Failure executing file: <myfile.py>
> 
>     how can i fix this? i'd like an n-by-m representation of my data as
>     an array which i can reference like a matrix in matlab. some of the
>     columns are floats, other are ints, and others are strings, so i
>     prefer to load the data into an array as a loop where i can cast the
>     strings appropriately, rather than use some built in io function for
>     reading tab-separated data.
> 
>     thank you very much.
> 
> 
> 
> ------------------------------------------------------------------------
> 
> 
> ------------------------------------------------------------------------
> 
> ------------------------------------------------------------------------------
> Open Source Business Conference (OSBC), March 24-25, 2009, San Francisco, CA
> -OSBC tackles the biggest issue in open source: Open Sourcing the Enterprise
> -Strategies to boost innovation and cut costs with open source participation
> -Receive a $600 discount off the registration fee with the source code: SFAD
> http://p.sf.net/sfu/XcvMzF8H
> 
> 
> ------------------------------------------------------------------------
> 
> _______________________________________________
> Matplotlib-users mailing list
> Mat...@li...
> https://lists.sourceforge.net/lists/listinfo/matplotlib-users

Horizontal bar is possible, although i'm not sure it fits your need.

http://matplotlib.sourceforge.net/examples/pylab_examples/barh_demo.html

Anyhow, with the current mpl, I guess it is much easy to make the
horizontal bars with rectangle patches (in case barh does not do what
you want), rather than trying to rotate the legend.

-JJ



On Sun, Mar 1, 2009 at 3:34 AM, Naoli <na...@tu...> wrote:
> Hi,
>
> thanks for the answer. That's what I thought.
> I'm using a bar chart plot to do a Gantt Project planner (please see
> example). Since it's not possible to draw horizontal bars, I plot a "normal"
> bar chart (vertical bars) and rotate the whole figure (please see attached
> file), this is why I need the legend to be 90°-rotated, so that it's in the
> right position afterall.
>
> Thanks for your time.
>
> Regards.
>
> N.
>
>
> 2009/3/1 Jae-Joon Lee <lee...@gm...>
>>
>> I don' think these is a straight forward way to rotate the legend as a
>> whole.
>> As a matter of fact, it is hardly possible with the current
>> implementation of the legend class. Could you explain why do you want
>> to have a rotated legend? An example figure (from other plotting
>> package) will be very helpful. If a rotated legend is desirable, I'll
>> work on to support it.
>>
>> Regards,
>>
>> -JJ
>>
>>
>> On Sat, Feb 28, 2009 at 4:10 AM, Naoli <na...@tu...> wrote:
>> > Hi guys,
>> >
>> > I was wondering if it's possible to rotate a legend ?
>> >
>> > Thanks for your help.
>> >
>> > Naoli
>> >
>> >
>> > ------------------------------------------------------------------------------
>> > Open Source Business Conference (OSBC), March 24-25, 2009, San
>> > Francisco, CA
>> > -OSBC tackles the biggest issue in open source: Open Sourcing the
>> > Enterprise
>> > -Strategies to boost innovation and cut costs with open source
>> > participation
>> > -Receive a $600 discount off the registration fee with the source code:
>> > SFAD
>> > http://p.sf.net/sfu/XcvMzF8H
>> > _______________________________________________
>> > Matplotlib-users mailing list
>> > Mat...@li...
>> > https://lists.sourceforge.net/lists/listinfo/matplotlib-users
>> >
>> >
>
>

hi all,

i am reading a set of tab-separated data from a file and i want to put it
into an array, and then plot some of the columns. i know the number of
columns ahead of time but not the number of rows. i load the array from the
file as follows, which seems to work:

data = []
for line in myfile:
  field1, field2, field3 = line.strip().split('\t')
  data.append([int(field1), int(field2), int(field3)])

i then convert it into an array as follows:

data = array(data)

i am able to reference the first column as follows:

data[:,0]

but if i try to plot the first column against the second as follows:

bar(data[:,0],data[:,1])

then i get the error:

/usr/lib64/python2.5/site-packages/matplotlib/units.pyc in
get_converter(self, x)
    128             converter = self.get(classx)
    129
--> 130         if converter is None and iterable(x):
    131             # if this is anything but an object array, we'll assume
    132             # there are no custom units

[repeated many times]

RuntimeError: maximum recursion depth exceeded
WARNING: Failure executing file: <myfile.py>

how can i fix this? i'd like an n-by-m representation of my data as an array
which i can reference like a matrix in matlab. some of the columns are
floats, other are ints, and others are strings, so i prefer to load the data
into an array as a loop where i can cast the strings appropriately, rather
than use some built in io function for reading tab-separated data.

thank you very much.

I don' think these is a straight forward way to rotate the legend as a whole.
As a matter of fact, it is hardly possible with the current
implementation of the legend class. Could you explain why do you want
to have a rotated legend? An example figure (from other plotting
package) will be very helpful. If a rotated legend is desirable, I'll
work on to support it.

Regards,

-JJ


On Sat, Feb 28, 2009 at 4:10 AM, Naoli <na...@tu...> wrote:
> Hi guys,
>
> I was wondering if it's possible to rotate a legend ?
>
> Thanks for your help.
>
> Naoli
>
> ------------------------------------------------------------------------------
> Open Source Business Conference (OSBC), March 24-25, 2009, San Francisco, CA
> -OSBC tackles the biggest issue in open source: Open Sourcing the Enterprise
> -Strategies to boost innovation and cut costs with open source participation
> -Receive a $600 discount off the registration fee with the source code: SFAD
> http://p.sf.net/sfu/XcvMzF8H
> _______________________________________________
> Matplotlib-users mailing list
> Mat...@li...
> https://lists.sourceforge.net/lists/listinfo/matplotlib-users
>
>

On Sat, Feb 28, 2009 at 9:59 PM, per freem <per...@gm...> wrote:
> hi Jae-Joon,
>
> thanks again, that makes sense.
>
> final question on this plot - i am trying to plot what we discussed but am
> getting very strange results.
>
> all i am trying to do is produce a scatter plot like:
>
> http://www2.warwick.ac.uk/fac/sci/moac/degrees/modules/ch923/r_introduction/scatter_plots/scatter.png
>
> where the key feature is that the first x tick and first y tick are not at
> the origin but spaced a bit away from it (and both equally spaced from it)
> and similarly for the last x and last y tick (which should be a bit before
> the end of the x and y axes.)
>
> if i plot on linux, i get the pdf attached as linux.pdf. if i plot on mac os
> x, i get the one attached as mac.pdf. aside from the differences in fonts,
> the linux pdf is lower quality where the ticks and the surrounding box of
> the plot have different thicknesses, which make the graph look strange.
> also, the xaxis labels are correctly spaced out from the ticks in the
> linux.pdf but not in the mac os x. overall, the mac pdf is of higher quality
> but the spacing is messed up.

Linux v.s Mac does not make much sense. What version of maplotlib do
you use in each platform? The rc file in two platform are identical?

They are both pdf files, it does not make much sense that one of them
has a higher quality. Looking at your two pdf files, the ticks in
linux.pdf looks a bit more fuzzy, but as far as I can see, it seems to
me as some antialising effect.

If you think the tick lines are too fuzzy, you 'd better increase the
width of tick lines.

 rcParams["lines.markeredgewidth"]=1. # default is 0.5

>
> how can i make the mac os x version space correctly, and make the linux
> version better looking?

I don't think there is nothing wrong with the mac version. But see
below. Again, if you think the tick lines are too fuzzy, increase the
line width.

>
> the code is simply:
>
> def axes_square(plot_handle):
>     plot_handle.axes.set_aspect(1/plot_handle.axes.get_data_ratio())
>
> fig = figure()
> rcParams['xtick.direction'] = 'out'
> rcParams['ytick.direction'] = 'out'
> s = subplot(1,3,1)
> x = arange(0,1,.01)
> y = arange(0,1,.01)+rand(100)
> axes_square(s)
>
> s = subplot(111)
> ax = plot(x, y, 'bo', markeredgecolor='blue', mfc='none')
>
> xlim(-.05,1,1.05)
> ylim(-.05,1,1.05)

Are the linux.pdf and mac.pdf created with a same script?
The linux one seems to have xrange of -0.05~1.05 while the mac one has
x-range of -0.05~1. And according to your script, it should be
-0.05~1.

What do you intend to do by "xlim(-0.05, 1, 1.05)"? Why you a third
arguments?  (I don't think you mean to set emit=True by this).

Maybe you just wanted "xlim(-0.05, 1.05)"?

-JJ

ps. I'm CC-ing this email to the list. Please CC to the list when you
reply so that the
conversation stays on the mailing list.

>
> savefig('filename.pdf')
>
> (so the only variability is the random vectors plotted)
>
> thanks again.
>>
>>
>> On Sat, Feb 28, 2009 at 6:22 PM, per freem <per...@gm...> wrote:
>> > hi,
>> >
>> > that code snippet works but it does not change the font. changing it to:
>> >
>> > def my_hash(self):
>> >    l = ['Courier',
>> >         self.get_style(),
>> >         self.get_variant(),
>> >         self.get_weight(),
>> >         self.get_stretch(),
>> >         self.get_size()]
>> >
>> >    return hash(repr(l))
>> > fp = FontProperties(family="Lucida Sans Typewriter")
>> > fp.__hash__ = my_hash
>> >
>> > runs, but does not change the font either.
>> >
>> > would be very interested if you have any other ideas? thanks.
>> >
>> >
>> >
>> >
>> > On Sat, Feb 28, 2009 at 6:04 PM, Jae-Joon Lee <lee...@gm...>
>> > wrote:
>> >>
>> >> On Sat, Feb 28, 2009 at 5:31 PM, per freem <per...@gm...> wrote:
>> >> > thank you for your reply. when i try either of the first suggestions
>> >> > about
>> >> > changing the fonts, i get the error:
>> >> >
>> >> > AttributeError: 'FontProperties' object has no attribute 'get_slant'
>> >> >
>> >> > any idea what this means?
>> >>
>> >> It seems that you're using v0.98.3 or before. See if following code
>> >> works.
>> >>
>> >> from matplotlib.font_manager import FontProperties
>> >>
>> >> def my_hash(self):
>> >>    l = [self.get_family(),
>> >>         self.get_style(),
>> >>         self.get_variant(),
>> >>         self.get_weight(),
>> >>         self.get_stretch(),
>> >>         self.get_size()]
>> >>
>> >>    return hash(repr(l))
>> >>
>> >> FontProperties.__hash__ = my_hash
>> >>
>> >>
>> >>
>> >> >
>> >> > also, i do not mind setting the position of each tickmark
>> >> > individually
>> >> > but i
>> >> > cannot find a way to do this -- could you please explain how this can
>> >> > be
>> >> > done?
>> >> >
>> >>
>> >>
>> >>
>> >> http://matplotlib.sourceforge.net/api/pyplot_api.html#matplotlib.pyplot.xticks
>> >>
>> >>
>> >> http://matplotlib.sourceforge.net/api/pyplot_api.html#matplotlib.pyplot.yticks
>> >>
>> >> -JJ
>> >>
>> >>
>> >> > thanks again.
>> >> >
>> >> >
>> >> > On Sat, Feb 28, 2009 at 5:19 PM, Jae-Joon Lee <lee...@gm...>
>> >> > wrote:
>> >> >>
>> >> >> > but it does not work. i tried similarly setting the font size
>> >> >> > (with
>> >> >> > set_size() or through rcParams) but it did not work either. how
>> >> >> > can i
>> >> >> > do
>> >> >> > this? i'd like to do this either on per axes basis, or for the
>> >> >> > entire
>> >> >> > figure.
>> >> >>
>> >> >> It seems that changing rcParams is not effective because of the way
>> >> >> how the font caching is done. Here is a little monkey patching to
>> >> >> change this behavior.
>> >> >>
>> >> >> from matplotlib.font_manager import FontProperties
>> >> >>
>> >> >> def my_hash(self):
>> >> >>    l = dict([(k, getattr(self, "get" + k)()) for k in
>> >> >> self.__dict__])
>> >> >>    return hash(repr(l))
>> >> >>
>> >> >> FontProperties.__hash__ = my_hash
>> >> >>
>> >> >>
>> >> >> With this code, changing rcParams will affect (most of) the text in
>> >> >> the
>> >> >> figure.
>> >> >>
>> >> >> As far as I know, you cannot have a default font properties on per
>> >> >> axes basis. You need to manually change the font properties of Text
>> >> >> artists in your interests.
>> >> >>
>> >> >> For example, to change the font properties of the xtick labels,
>> >> >>
>> >> >> fp = FontProperties(family="Lucida Sans Typewriter")
>> >> >> ax = gca()
>> >> >> for t in ax.get_xticklabels():
>> >> >>    t.set_fontproperties(fp)
>> >> >>
>> >> >>
>> >> >> >
>> >> >> > second, how can i make it so axes labels do not overlap? in many
>> >> >> > plots,
>> >> >> > including ones in the gallery, you see the labels at the origin of
>> >> >> > plots
>> >> >> > get
>> >> >> > too close to each other. (i.e. the 0.0 of x-axis and 0.0 of
>> >> >> > y-axis) -
>> >> >> > how
>> >> >> > can you prevent this from happening?
>> >> >> >
>> >> >>
>> >> >> I don't think there is a smart way to prevent it other than manually
>> >> >> changing the tick positions. Other may have better ideas.
>> >> >>
>> >> >> -JJ
>> >> >
>> >> >
>> >
>> >
>
>