matplotlib-users Mailing List for matplotlib

Yep. I would like to pass in a list of lists, where each sublist (or array) 
describes a boxplot to plot.
Meanwhile, i've been having fun with histograms. The Y axis labels are a 
pain. I think defaulting to scientific notation, as matplotlib frequently 
does, is annoying...
___
Sent with SnapperMail
www.snappermail.com
...... Original Message .......
On 2006年12月16日 11:02:28 -1000 "Eric Firing" <ef...@ha...> wrote:
>Pierre GM wrote:
>>> And for the work I'm doing, I have a
>>> different number of observations and data points on different days,
>>> so it's a pain that the current boxplot infrastructure expects all of
>>> the boxes to be in a single array. Hence my questions.
>> 
>> Ah OK, now I get it. Sorry for being a bit slow today.
>> So yes, there's a problem here. The sequence is transformed into an 
array 
>> with 'asarray', which obviously won't work if you have different sizes 
of 
>> dataset.
>> 
>> But there's a trick, tested on numpy:
>> 
>> import numpy
>> set1 = (rand(50)+1) * 100
>> set2 = (rand(25)+2) * 100
>> data = array([set1,set2], dtype=numpy.object_)
>> boxplot(data,positions=[732659.,732660.])
>> 
>> The trick here is to force the data as objects, and not as floats. Your array 
>> is in fact an array of arrays of different sizes. That's enough to fool 
>> asarray.
>
>It sounds like the real problem is that the initial use of asarray in 
>boxplot is a bug--it should transparently support an object array, as 
>you suggest (but numpy only), or an ordinary array, *or* a list or tuple 
>of data vectors, and all this should be clear in the docstring and the 
>example. Correct? I hope so, because I have made the change in svn. 
>(Well, maybe not enough changes to the example yet.)
>
>Eric
>

On Saturday 16 December 2006 16:02, Eric Firing wrote:
> It sounds like the real problem is that the initial use of asarray in
> boxplot is a bug--it should transparently support an object array, as
> you suggest (but numpy only), or an ordinary array, *or* a list or tuple
> of data vectors, and all this should be clear in the docstring and the
> example. Correct? I hope so, because I have made the change in svn.
> (Well, maybe not enough changes to the example yet.)
Fully agreed. There's no need for the transformation to array: the only 
variable which is array-related is the number of columns (viz, the number of 
boxes to plot).
OK, there's also that:
for i,pos in enumerate(positions):
 d = x[:,i]
But it'd be as easy to enumerate the array (after a first transpose, if its a 
regular ndarray...)

Pierre GM wrote:
>> And for the work I'm doing, I have a
>> different number of observations and data points on different days,
>> so it's a pain that the current boxplot infrastructure expects all of
>> the boxes to be in a single array. Hence my questions.
> 
> Ah OK, now I get it. Sorry for being a bit slow today.
> So yes, there's a problem here. The sequence is transformed into an array 
> with 'asarray', which obviously won't work if you have different sizes of 
> dataset.
> 
> But there's a trick, tested on numpy:
> 
> import numpy
> set1 = (rand(50)+1) * 100
> set2 = (rand(25)+2) * 100
> data = array([set1,set2], dtype=numpy.object_)
> boxplot(data,positions=[732659.,732660.])
> 
> The trick here is to force the data as objects, and not as floats. Your array 
> is in fact an array of arrays of different sizes. That's enough to fool 
> asarray.
It sounds like the real problem is that the initial use of asarray in 
boxplot is a bug--it should transparently support an object array, as 
you suggest (but numpy only), or an ordinary array, *or* a list or tuple 
of data vectors, and all this should be clear in the docstring and the 
example. Correct? I hope so, because I have made the change in svn. 
(Well, maybe not enough changes to the example yet.)
Eric

Sergio,
I think the line you see comes from plotting two shape on top one another. You 
may want to redefine your polygon:
the easier would be something like:
tt = concatenate((t,t[::-1]))
vtx = concatenate((s3,s2[::-1]))
fill(tt, vtx, '#e9ccce', linewidth=0)
With the first command, you build a series of x coordinates from 0 to 1.01 and 
back to 0 (that's the t[::-1])
with second, you build the y ccordinates sequence, by taking the corresponindg 
point of s3, and then from s2 in reverse order. 
Now, your polygon is fully defined.

Hi,
I am a new user of matplotlib.
First of all I would like to say that the people working on matplotlib is
doing a great job.
Well, I have tried to use the 'fill', but I cannot get rid of the lines.
I have tried to use 'linewidth=0' at the end, but it did not work.
Could somebody, please, tell me if I am doing something wrong ?
Thank you.
Sergio.
------------------------------------------------------------------
#!/usr/bin/env python
from pylab import *
from matplotlib.patches import Polygon
t = arange(0.0, 1.0+0.01, 0.01)
s1 = t
s2 = t*t
s3 = t*t*t
text(0.50, 0.4,
 r"$d(f_2,f_3)$", horizontalalignment='center',
 color= '#8f0100', fontsize=30)
xticks(size=20)
yticks(size=20)
fill(t, s3, '#e9ccce', linewidth=0)
fill(t, s2, 'w', linewidth=0)
savefig('diamxn2.eps')

Hi,
I am a new user of matplotlib.
First of all I would like to say that the people working on matplotlib is
doing a great job.
Well, I have tried to use the 'fill', but I cannot get rid of the lines.
I have tried to use 'linewidth=0' at the end, but it did not work.
Could somebody, please, tell me if I am doing something wrong ?
Thank you.
Sergio.
------------------------------------------------------------------
#!/usr/bin/env python
from pylab import *
from matplotlib.patches import Polygon
t = arange(0.0, 1.0+0.01, 0.01)
s1 = t
s2 = t*t
s3 = t*t*t
text(0.50, 0.4,
 r"$d(f_2,f_3)$", horizontalalignment='center',
 color= '#8f0100', fontsize=30)
xticks(size=20)
yticks(size=20)
fill(t, s3, '#e9ccce', linewidth=0)
fill(t, s2, 'w', linewidth=0)
savefig('diamxn2.eps')

On Saturday 16 December 2006 13:47, Simson Garfinkel wrote:
> > I apologize if I offended anyone, this was really not my intention
> > at all.
>
> Oh, I was never offended.
Thanks a lot for your patience nevertheless.
> I think that it's possible to get impressive results in a few hours,
> but not become a pro. 
I see that as an advantage: you can never get bored, there's always something 
new to learn (I should have listened to my mom and spend less time in front 
of the TV as a kid). But I admit it can get a bit frustrating, when you 
eventually find out that there was a really simple solution to a problem, 
alas undocumented and buried in the code (for example, the xaxis_date thingie 
which doesn't do much).
> I've been using matplotlib for something like 
> 20-40 hours right now. I've done some neat things, but have had some
> problems.
That's when the mailing list becomes really useful.
> I've also written a neat pre-processor that allows you to embed
> python and matplotlib code in LaTeX, so you don't need to have it all
> spread out. And you can populate the results from SQL queries, right
> there in the LaTeX. It makes paper writing much easier.
Oh, that sounds great ! Could you post it somewhere ? I'm sure it'd be quite 
useful (I do have a need for it myself...)

>
> I apologize if I offended anyone, this was really not my intention 
> at all.
Oh, I was never offended.
> My
> point was that after only a few hours, it is indeed possible to get
> impressive results and become a real MPL pro.
I think that it's possible to get impressive results in a few hours, 
but not become a pro. I've been using matplotlib for something like 
20-40 hours right now. I've done some neat things, but have had some 
problems.
I've also written a neat pre-processor that allows you to embed 
python and matplotlib code in LaTeX, so you don't need to have it all 
spread out. And you can populate the results from SQL queries, right 
there in the LaTeX. It makes paper writing much easier.

On Saturday 16 December 2006 13:01, David Chin wrote:
> On 12/16/06, Simson Garfinkel <si...@ac...> wrote:
> > Thanks. I have roughly 30 years of programming experience and know
> > something like 20+ languages. The learning curve here is steep, and I
> > think that a lot could be done to make it less steep.
>
> I agree. Pierre, maybe you should google Simson's name here
> before taking a condescending tone.
I apologize if I offended anyone, this was really not my intention at all. My 
point was that after only a few hours, it is indeed possible to get 
impressive results and become a real MPL pro.
> Anyway, I agree with the other suggestion that a lot of the stuff here
> should be wikied. Now, Pierre said that one can edit the wiki, but when
> I tried it ( http://scipy.org/Cookbook/Matplotlib?action=edit ), I get an
> error saying "You are not allowed to edit this page." So, may I suggest
> that the Wiki be opened up for editing? Information flows much more
> quickly if everyone can contribute.
Well, I think you have to register before being able to edit pages. There's a 
login button on the top right of your screen.
But I agree that it's not intuitive (I have to poke around for a while to find 
it) (and then remember my login name).

Hi,
Each time I'm working on C++ codes using vector or valarray, I would
like to be able to plot them.
The problem is that there is no straitforward way to do that in C++.
My goal is not to code a QT or GTK application but only to be able to
plot 1D and 2D things from one given large C++ code without having to
add lots of lines of codes in my code (let say it is intend to be used
in debug phase).
Questions :
Is there a way to call pylab plot and imshow from a C++ code ?
In this case, I do not care if we have to copy the array and it can be sl=
ow.
It would be a so nice feature to debug C++ image processing codes.
Any example of code is welcome even they are not calling matplotlib but
anthing else in python.
Xavier.
ps : In my codes, 2D images are stored as in a class derived from
valarray (1D array) adding the size of the image along the 2 directions
as private members.
--=20
############################################
Xavier Gnata
CRAL - Observatoire de Lyon
9, avenue Charles Andr=E9
69561 Saint Genis Laval cedex
Phone: +33 4 78 86 85 28
Fax: +33 4 78 86 83 86
E-mail: gn...@ob...
############################################=20

> And for the work I'm doing, I have a
> different number of observations and data points on different days,
> so it's a pain that the current boxplot infrastructure expects all of
> the boxes to be in a single array. Hence my questions.
Ah OK, now I get it. Sorry for being a bit slow today.
So yes, there's a problem here. The sequence is transformed into an array 
with 'asarray', which obviously won't work if you have different sizes of 
dataset.
But there's a trick, tested on numpy:
import numpy
set1 = (rand(50)+1) * 100
set2 = (rand(25)+2) * 100
data = array([set1,set2], dtype=numpy.object_)
boxplot(data,positions=[732659.,732660.])
The trick here is to force the data as objects, and not as floats. Your array 
is in fact an array of arrays of different sizes. That's enough to fool 
asarray.

On Dec 16, 2006, at 1:01 PM, Pierre GM wrote:
>> Hi, Pierre. There's a lot of assumptions here.
> Indeed, and I apologize
>
>> I sort of know what numarray, Numeric and numpy are, but I don't care
>> all that much. I'm just interested in matplotlib for the plotting.
> Well, matplotlib relies on some packages to handle data arrays. 
> It's probably
> a good idea to check what you have below the hood, just in case 
> something
> goes wrong. It'll make things easier to help you.
Thanks. In fact, after I read through the matplotlib tutorial and 
user's manual, I downloaded and read the numpy manual too. It's a 
nice package. I especially liked the line about numpy not making 
things possible that weren't possible before, but that it just allows 
things to be finished before we are dead.
Overall, the package is really impressive. I don't like the way that 
arrays aren't resizable, though. And for the work I'm doing, I have a 
different number of observations and data points on different days, 
so it's a pain that the current boxplot infrastructure expects all of 
the boxes to be in a single array. Hence my questions.
>
>> The point of my email (and the questions below) is that the entire
>> scipy.org website is written for people who already know what's going
>> on. If you want to expand the user community (and simultaneously cut
>> down on questions) you need to make things dramatically more
>> understandable than they currently are.
>
> That's true, and I guess that's the reason why there's a wiki. 
> Please feel
> free to add your comments and suggestions.
Thanks. I will.
>
>> A related problem is the MoinMoin wiki. It is not a widely used wiki
>> and it has many usability problems. It is dramatically harder to use
>> than, say, mediawiki.
>
> But it happens to be the solution that has been chosen.
Yep.
>

> I want multiple boxes on a single plot, with one box per day. Take a
> look at how I've done it with just plot() and some error bars...
I'm still not sure I understand where the problem is:
You want several boxes in a plot ?
Something along the lines of what I already sent you ?
> boxplot([set1, set2],positions=[732659,732660])
Then once again, just construct of sequences of sets, a sequence of dates (as 
ordinals), and use something like:
boxplot(setsequence, positions=datesequence).
Alternatively, follow the boxplot_demo example.
Then, you'll have to use the trick I sent earlier
gca().set_xticklabels([num2date(x).strftime(fmt) for x in 
gca().get_xticklabels()])

On 12/16/06, Simson Garfinkel <si...@ac...> wrote:
>
> On Dec 16, 2006, at 12:02 PM, Pierre GM wrote:
> > I know, the learning curve is a bit steep at first, but soon you'll
> > be a real
> > pro.
>
> Thanks. I have roughly 30 years of programming experience and know
> something like 20+ languages. The learning curve here is steep, and I
> think that a lot could be done to make it less steep.
I agree. Pierre, maybe you should google Simson's name here
before taking a condescending tone.
Anyway, I agree with the other suggestion that a lot of the stuff here
should be wikied. Now, Pierre said that one can edit the wiki, but when
I tried it ( http://scipy.org/Cookbook/Matplotlib?action=edit ), I get an
error saying "You are not allowed to edit this page." So, may I suggest
that the Wiki be opened up for editing? Information flows much more
quickly if everyone can contribute.
Cheers,
 Dave

> Hi, Pierre. There's a lot of assumptions here.
Indeed, and I apologize
> I sort of know what numarray, Numeric and numpy are, but I don't care
> all that much. I'm just interested in matplotlib for the plotting.
Well, matplotlib relies on some packages to handle data arrays. It's probably 
a good idea to check what you have below the hood, just in case something 
goes wrong. It'll make things easier to help you.
> The point of my email (and the questions below) is that the entire
> scipy.org website is written for people who already know what's going
> on. If you want to expand the user community (and simultaneously cut
> down on questions) you need to make things dramatically more
> understandable than they currently are.
That's true, and I guess that's the reason why there's a wiki. Please feel 
free to add your comments and suggestions.
> A related problem is the MoinMoin wiki. It is not a widely used wiki
> and it has many usability problems. It is dramatically harder to use
> than, say, mediawiki.
But it happens to be the solution that has been chosen. 

I agree. It may be common in matlab, but it really doesn't belong in 
python.
On Dec 16, 2006, at 12:50 PM, Eric Firing wrote:
>
>>> BTW, this whole subplot(ijk) instead of subplot(i,j,k) notation is
>>> really, really confusing to me...
>> Don't get overwhelmed. ijk is a shortcut for (i, j, k), that works 
>> well if you're working with less than 10 plots in either direction.
>
> It is a holdover from the early days of Matlab. It makes mpl more 
> Matlab-like (for better or worse) and saves 2-4 keystrokes. 
> Personally, I don't like it and would be inclined to discourage it 
> in mpl.
>
> Eric
>

>> BTW, this whole subplot(ijk) instead of subplot(i,j,k) notation is
>> really, really confusing to me...
> 
> Don't get overwhelmed. ijk is a shortcut for (i, j, k), that works well if 
> you're working with less than 10 plots in either direction.
It is a holdover from the early days of Matlab. It makes mpl more 
Matlab-like (for better or worse) and saves 2-4 keystrokes. Personally, 
I don't like it and would be inclined to discourage it in mpl.
Eric

On Dec 16, 2006, at 12:02 PM, Pierre GM wrote:
>
>> I'd love to speak python to it. But it's harder when all of the
>> examples are in matlab...
>
> :)
> Well, please have a look to pythonic_matplotlib.py in your examples 
> folder.
>
>>> fig = figure()
>>> ax1 = fig.add_subplot(121)
>>> ax2=fig.add_subplot(122)
>>
>> Hm. I'll need to figure out why these two subplots appear on the same
>> axis.
>
> What do you mean ? You want two plots on a figure, or two figures?
> You want one plot in the topleft corner, one in hte bottom right ? 
> You can do
> that as well, just tell matplotlib where to plot the axes (a bit of
> terminology here: an axes is a box in your figure, in which you 
> will draw a
> subplot).
I want multiple boxes on a single plot, with one box per day. Take a 
look at how I've done it with just plot() and some error bars...

On Dec 16, 2006, at 11:58 AM, Pierre GM wrote:
>> I'm very confused by the wiki in general. I click on "wiki" and it 
>> takes me
>> to something that doesn't obviously have anything to do with 
>> matplotlib...
>
> Well, it does say: matplotlib cookbook.
>
>> Like, what's scipy.org? Is it a company? Who is EnThought?
> Oh.
> What are you using to manipulate arrays ? numarray, Numeric, or 
> numpy ?
> Assuming that you use numpy, then you must know what scipy is, 
> right ? If
> not, well, br
> scipy.org is just a central site for numpy/scipy-related information.
> Enthought is a private company that works extensively with Python 
> and numpy
> in particular, and that hosts the site.
Hi, Pierre. There's a lot of assumptions here.
I do most of my own manipulations in SQL and with custom-written 
programs in C, C++, and Python.
I sort of know what numarray, Numeric and numpy are, but I don't care 
all that much. I'm just interested in matplotlib for the plotting.
The point of my email (and the questions below) is that the entire 
scipy.org website is written for people who already know what's going 
on. If you want to expand the user community (and simultaneously cut 
down on questions) you need to make things dramatically more 
understandable than they currently are.
A related problem is the MoinMoin wiki. It is not a widely used wiki 
and it has many usability problems. It is dramatically harder to use 
than, say, mediawiki.
>
>> Amy I allowed to contribute to the wiki? It doesn't look like it. 
>> The whole
>> thing is not very friendly.
>
> It's a wiki, so yes, you can contribute. No, it's not especially 
> friendly, in
> the sense that nobody's here to hold your hand.
>
>> And what is this whole MoinMoin thing?
> "MoinMoin is a Python WikiClone, based on PikiPiki. "
> This converstaion is really becoming surrealistic.
>

> I'd love to speak python to it. But it's harder when all of the
> examples are in matlab...
:)
Well, please have a look to pythonic_matplotlib.py in your examples folder.
> > fig = figure()
> > ax1 = fig.add_subplot(121)
> > ax2=fig.add_subplot(122)
>
> Hm. I'll need to figure out why these two subplots appear on the same
> axis.
What do you mean ? You want two plots on a figure, or two figures?
You want one plot in the topleft corner, one in hte bottom right ? You can do 
that as well, just tell matplotlib where to plot the axes (a bit of 
terminology here: an axes is a box in your figure, in which you will draw a 
subplot).
> BTW, this whole subplot(ijk) instead of subplot(i,j,k) notation is
> really, really confusing to me...
Don't get overwhelmed. ijk is a shortcut for (i, j, k), that works well if 
you're working with less than 10 plots in either direction.
I know, the learning curve is a bit steep at first, but soon you'll be a real 
pro.

> I'm very confused by the wiki in general. I click on "wiki" and it takes me
> to something that doesn't obviously have anything to do with matplotlib...
Well, it does say: matplotlib cookbook.
> Like, what's scipy.org? Is it a company? Who is EnThought?
Oh.
What are you using to manipulate arrays ? numarray, Numeric, or numpy ? 
Assuming that you use numpy, then you must know what scipy is, right ? If 
not, well, br
scipy.org is just a central site for numpy/scipy-related information. 
Enthought is a private company that works extensively with Python and numpy 
in particular, and that hosts the site.
> Amy I allowed to contribute to the wiki? It doesn't look like it. The whole
> thing is not very friendly.
It's a wiki, so yes, you can contribute. No, it's not especially friendly, in 
the sense that nobody's here to hold your hand.
> And what is this whole MoinMoin thing?
"MoinMoin is a Python WikiClone, based on PikiPiki. "
This converstaion is really becoming surrealistic.

>
>
>> Now, how do I get two boxplots on the same plot?
>
> Well, just draw two axes.
> Simson, now that you're more experienced with matplotlib, you 
> should really
> start speaking python to it.
I'd love to speak python to it. But it's harder when all of the 
examples are in matlab...
>
> fig = figure()
> ax1 = fig.add_subplot(121)
> ax2=fig.add_subplot(122)
Hm. I'll need to figure out why these two subplots appear on the same 
axis.
BTW, this whole subplot(ijk) instead of subplot(i,j,k) notation is 
really, really confusing to me...
>
> ax1.boxplot([set1, set2],positions=[732659,732660])
> ax2.boxplot([set2, set1],positions=[732659,732660])
> ax1.set_xticklabels([num2date(x).strftime(timefmt) for x in 
> ax1.get_xticks()])
> ax2.set_xticklabels([num2date(x).strftime(timefmt) for x in 
> ax2.get_xticks()])
>
>

On Friday 15 December 2006 21:07, Simson Garfinkel wrote:
> Hm. thanks for the info. But it's not perfect... I get times in my
> formats, but not the dates. Here is the sample code:
Yeah, I agree, the situation is far from ideal. Besides, it turns out that 
there's no deep magic behind have_dates, which is just a way to tell the axis 
to use AutoDateFormatter. Which we don't need.
So:
Plot your boxeds with the positions flag:
> boxplot([set1, set2],positions=[732659,732660])
> ax = gca()
Then use num2date:
timefmt = '%b-%d-%Y'
gca().set_xticklabels([num2date(x).strftime(timefmt) for x in 
gca().get_xticks()])
> Now, how do I get two boxplots on the same plot?
Well, just draw two axes.
Simson, now that you're more experienced with matplotlib, you should really 
start speaking python to it.
fig = figure()
ax1 = fig.add_subplot(121)
ax2=fig.add_subplot(122)
ax1.boxplot([set1, set2],positions=[732659,732660])
ax2.boxplot([set2, set1],positions=[732659,732660])
ax1.set_xticklabels([num2date(x).strftime(timefmt) for x in ax1.get_xticks()])
ax2.set_xticklabels([num2date(x).strftime(timefmt) for x in ax2.get_xticks()])

It would be useful if grid had two options:
	drawHorizontalLines = True
	drawVerticalLines = True
This way, people who look for suppressing the X or Y lines would find 
it in the logical location.

----- Original Message ----- 
From: "Angus McMorland" <am...@gm...>
> We really shoud wiki more of these email discussions as they come
> along. It's so much easier to search there, since things are in some
> sort of logical arrangement.
I agree.
I'm very confused by the wiki in general. I click on "wiki" and it takes me 
to something that doesn't obviously have anything to do with matplotlib... 
Like, what's scipy.org? Is it a company? Who is EnThought?
Amy I allowed to contribute to the wiki? It doesn't look like it. The whole 
thing is not very friendly.
And what is this whole MoinMoin thing?