matplotlib-users Mailing List for matplotlib

On Sun, 10 Feb 2008, David Trémouilles apparently wrote:
> I have a slightly different objective: I just want to 
> remove outliers 

Do you just want to filter out the outliers?

    newdata = [datum for datum in data if not isoutlier(datum)]

You can define ``isoutlier`` to return True for outliers
in your data.

Apologies if this proves OT.

fwiw,
Alan Isaac

I would like to be able to use Cambria font  (http://en.wikipedia.org/wiki/Cambria_(typeface)) for all text on my charts. 

I am adding these charts to a MS Word 2007 document written in the same font.

I tried to add Cambria as the fist string in rcsetup.py | defaultParams | 'font.serif'.  That didn't work.  Can I use this font for the charts?

On Sunday 10 February 2008 13:23:13 David Trémouilles wrote:
> Thank you very much Pierre!
> You made me discover boolean index (numpy is fantastic !)
> In the mean time, I now understand the purpose of maskedarray that I
> totally missed at a first sight.

You're quite welcome. Masked arrays are great when you need a way to flag 
invalid or missing data. For simpler cases, boolean indexing can be faster 
and easier to understand.
And now, for a shameless plug: if you work with series indexed with time, you 
might be interested in the timeseries package (available as a scikit in 
http://svn.scipy.org/svn/scikits/trunk/timeseries/). The package relies on 
the new numpy.ma package.

Thank you very much Pierre!
You made me discover boolean index (numpy is fantastic !)
In the mean time, I now understand the purpose of maskedarray that I 
totally missed at a first sight.

Thanks to all of you,

David

Pierre GM a écrit :
> On Sunday 10 February 2008 12:40:38 David Trémouilles wrote:
> 
>> I have a slightly different objective: I just want to remove outliers
>> from my curves. I think I will still play with maskedarray and used the
>> compressed() function before 'sending' to matplotlib.
>> Any comments on that, any other idea?
> 
> So, you have two arrays x and y, with missing values in y that you don't want 
> to plot ?
> Assuming that your arrays are 1D, you can try something like:
> plot(x[logical_not(y.mask)], y.compressed())
> in order to ensure that the x and y to be plotted have the same size.
> 
> Note that in this simple case, you don't need masked arrays, you just want to 
> plot point satisfying a given condition, right ?
> So:
> condition = (y>=min_value) & (y<= max_value)
> plot(x[condition],y[condition])
> will give the same results.
> 
> -------------------------------------------------------------------------
> This SF.net email is sponsored by: Microsoft
> Defy all challenges. Microsoft(R) Visual Studio 2008.
> http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/
> _______________________________________________
> Matplotlib-users mailing list
> Mat...@li...
> https://lists.sourceforge.net/lists/listinfo/matplotlib-users

On Sunday 10 February 2008 12:40:38 David Trémouilles wrote:

> I have a slightly different objective: I just want to remove outliers
> from my curves. I think I will still play with maskedarray and used the
> compressed() function before 'sending' to matplotlib.
> Any comments on that, any other idea?

So, you have two arrays x and y, with missing values in y that you don't want 
to plot ?
Assuming that your arrays are 1D, you can try something like:
plot(x[logical_not(y.mask)], y.compressed())
in order to ensure that the x and y to be plotted have the same size.

Note that in this simple case, you don't need masked arrays, you just want to 
plot point satisfying a given condition, right ?
So:
condition = (y>=min_value) & (y<= max_value)
plot(x[condition],y[condition])
will give the same results.

Thanks Jeff, I think now I get the purpose of maskedearray the way it is 
used in matplotlib.

I have a slightly different objective: I just want to remove outliers 
from my curves. I think I will still play with maskedarray and used the 
compressed() function before 'sending' to matplotlib.
Any comments on that, any other idea?

Thanks,

David

Jeff Whitaker a écrit :
> David Trémouilles wrote:
>> Hi,
>>
>>  I've just start playing with maskedarray (the new implementation) 
>> using fresh svn matplotib (0_91 maintenance).
>> Plotting masked array does not behave as I would have expected.
>> Indeed when drawing a "line graph" the masked walues interrupted the 
>> line (see attach example).
> 
> David:  Yes, this is the correct behavior.  The masked values are 
> treated as missing data.  No attempt is made to fill, or interpolate, 
> the missing data.
>> I would prefer to see a continues line...
> 
> Then you should interpolate the missing values yourself.  I think it 
> would be unwise for matplotlib to guess how you might want to do that.
> 
> -Jeff
>> Is it the expected behavior? Is there a way to change it?
>>
>> Thanks in advance,
>>
>> David
>> ------------------------------------------------------------------------
>>
>> ------------------------------------------------------------------------
>>
>> -------------------------------------------------------------------------
>> This SF.net email is sponsored by: Microsoft
>> Defy all challenges. Microsoft(R) Visual Studio 2008.
>> http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/
>> ------------------------------------------------------------------------
>>
>> _______________________________________________
>> Matplotlib-users mailing list
>> Mat...@li...
>> https://lists.sourceforge.net/lists/listinfo/matplotlib-users
> 
> 

On Feb 10, 2008 9:43 AM, John Hunter <jd...@gm...> wrote:

> > I want to create a scatter plot onto a background image. Anybody could help
> > me?Thank you!
>
> The mri_demo.py example in the matplotlib/examples in the src
> distribution illustrates this.

Sorry -- wrong example.  The example I was thinking of is
examples/image_demo2.py, also at

http://matplotlib.sf.net/examples/image_demo2.py

but if you want to run it you will need to get the matplotlib src
directory which also has the data file for the CT image

JDH

2008/2/9 He Jibo <he...@gm...>:
> Hi, Everyone,
>
> I want to create a scatter plot onto a background image. Anybody could help
> me?Thank you!

The mri_demo.py example in the matplotlib/examples in the src
distribution illustrates this.

David Trémouilles wrote:
> Hi,
>
>  I've just start playing with maskedarray (the new implementation) 
> using fresh svn matplotib (0_91 maintenance).
> Plotting masked array does not behave as I would have expected.
> Indeed when drawing a "line graph" the masked walues interrupted the 
> line (see attach example).

David:  Yes, this is the correct behavior.  The masked values are 
treated as missing data.  No attempt is made to fill, or interpolate, 
the missing data.
> I would prefer to see a continues line...

Then you should interpolate the missing values yourself.  I think it 
would be unwise for matplotlib to guess how you might want to do that.

-Jeff
> Is it the expected behavior? Is there a way to change it?
>
> Thanks in advance,
>
> David
> ------------------------------------------------------------------------
>
> ------------------------------------------------------------------------
>
> -------------------------------------------------------------------------
> This SF.net email is sponsored by: Microsoft
> Defy all challenges. Microsoft(R) Visual Studio 2008.
> http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/
> ------------------------------------------------------------------------
>
> _______________________________________________
> Matplotlib-users mailing list
> Mat...@li...
> https://lists.sourceforge.net/lists/listinfo/matplotlib-users


-- 
Jeffrey S. Whitaker         Phone : (303)497-6313
NOAA/OAR/CDC  R/PSD1        FAX   : (303)497-6449
325 Broadway                Boulder, CO, USA 80305-3328

Hi,

  I've just start playing with maskedarray (the new implementation) 
using fresh svn matplotib (0_91 maintenance).
Plotting masked array does not behave as I would have expected.
Indeed when drawing a "line graph" the masked walues interrupted the 
line (see attach example). I would prefer to see a continues line...
Is it the expected behavior? Is there a way to change it?

Thanks in advance,

David

Hi He


2008/2/10 He Jibo <he...@gm...>:
> Hi, Arnar,
>
> Thank you for your great help!This is what I need! But this code has some
> warnings, could you please help me debug it? Thank you !
>
> the first error is :
> Traceback (most recent call last):
>   File "D:/Program Files/python 2.4.3/py2exe
> code/fixation_distribution_background.py", line 10, in -toplevel-
>
>     xy = p.loadtxt(xy_fn)
> AttributeError: 'module' object has no attribute 'loadtxt'

loadtxt is from the numpy interface, so I assume you are not using a
version of matplotlib with numpy. load() should do the job as well.
The point is to get an array (xy) of shape (n_points x 2), see below.

> I guess 'loadtxt' should be 'load'. After I changed this, another error
> came. It is:
>
> Traceback (most recent call last):
>   File "D:/Program Files/python 2.4.3/py2exe
> code/fixation_distribution_background.py", line 11, in -toplevel-
>     xy_normed = xy/[[1024./im.shape[1], 768./im.shape[0]]]
>  ValueError: shape mismatch: objects cannot be broadcast to a single shape

If the array xy has shape (8,2) this broadcast should be fine. Again,
this could be an issue with your version of Numeric/Numarray/(old
numpy). I dont know these old libraries well enough to say for sure. I
think at least Numeric uses same type of broadcasting as numpy.

So, what kind of numeric library are you using and what is the shape
of xy after you have loaded the fixation-textfile?


Arnar

PS:
For reference my versions are as follows:
numpy.__version__   :  '1.0.5.dev'
matplotlib.__version__  : '0.91.2'

Broadcasting explained: http://www.scipy.org/EricsBroadcastingDoc

Hi Jibo,

I'm not sure of your reasons for wanting to do this, but you might find
the psychopy package of interest:
http://www.psychopy.org/

Gary R.

He Jibo wrote:
> Hi, Everyone,
>  
> I want to create a scatter plot onto a background image. Anybody could 
> help me?Thank you!
>  
> The background.PNG is shown full screen with a resolution of 1024*768. 
> The data in the fixation_xy.txt is the coordinates of eye-movement data, 
> first column for X axis, second column for Y axis. I wish to do a 
> scatter plot with the data onto background.PNG. Please give me a helping 
> hand how could I do this with matplotlib.
>  
> Thank you !
>  
> Jibo
>  
> 
> 
> -- 
> Best Regards,
> 
> 　　　　　　　　 He Jibo
> 　　　　　　　ji...@cy... 
> <mailto:ji...@cy...>
> he...@gm... <mailto:he...@gm...>

> The background.PNG is shown full screen with a resolution of 1024*768. The
> data in the fixation_xy.txt is the coordinates of eye-movement data, first
> column for X axis, second column for Y axis. I wish to do a scatter plot
> with the data onto background.PNG. Please give me a helping hand how could I
> do this with matplotlib.

!! I am assuming xy-fixation (screen coordinates) values are given
with respect to a a lower left origin. (see line [*] in code)

import pylab as p

im_fn = "background.PNG"
xy_fn = "fixation_xy.txt"
s = 100 # marker size
c = 'r'  # marker color
marker = 's' # marker type

im = p.imread(im_fn)
xy = p.loadtxt(xy_fn)
xy_normed = xy/[[1024./im.shape[1], 768./im.shape[0]]]

p.imshow(im, origin='lower') # [*]
p.scatter(xy_normed[:,0], xy_normed[:,1], s=s, c=c, marker=marker)
p.show()


Arnar