Thread: [Matplotlib-users] plot cdf

SourceForge Headquarters 225 Broadway Suite 1600 San Diego, CA 92101 +1 (858) 422-6466

Is there a standard function or practice for
plotting the CDF of a series?  (I am aware
of the output of hist.)

Thank you,
Alan Isaac

On Sep 26, 2007, at 5:09 PM, Alan Isaac wrote:

> Is there a standard function or practice for
> plotting the CDF of a series?  (I am aware
> of the output of hist.)

Without really knowing what CDF is (I am assuming
it is Cumulative Density Fucntion or something similar).
I would suggest taking a look at numpy and the histogram
and cumsum functions therein.

Cheers
   Tommy

Alan Isaac wrote:
> Is there a standard function or practice for
> plotting the CDF of a series?  (I am aware
> of the output of hist.)

import numpy as np
from matplotlib import pylab

x = ... # whatever
n = len(x)
x2 = np.repeat(x, 2)
y2 = np.hstack([0.0, np.repeat(np.arange(1,n) / float(n), 2), 1.0])
pylab.plot(x2, y2)

-- 
Robert Kern

"I have come to believe that the whole world is an enigma, a harmless enigma
 that is made terrible by our own mad attempt to interpret it as though it had
 an underlying truth."
  -- Umberto Eco

On Wed, 26 Sep 2007, Robert Kern apparently wrote:
> n = len(x)
> x2 = np.repeat(x, 2)
> y2 = np.hstack([0.0, np.repeat(np.arange(1,n) / float(n), 2), 1.0])
> pylab.plot(x2, y2)

OK, that's pretty slick.
I did not think about ``repeat``.
Thanks,
Alan

After thinking it over, I did not go for
Robert or David's cool numpy tricks, but
I'll append a simple object in case someone
else wants to do more.

Cheers,
Alan Isaac

class EmpiricalCDF(object):
    '''Empirical cdf.
    First point will be (xmin,0).
    Last point will be (xmax,1).
    '''
    def __init__(self, data, sortdata=True):
        if sortdata:
            data = N.sort(data)
        self.data = data
        self.nobs = len(data)
    def gen_xp(self):
        data, nobs = self.data, self.nobs
        prob = N.linspace(0, 1, nobs+1)
        xsteps = ( data[(idx)//2] for idx in xrange(2*nobs) )
        psteps = ( prob[(idx+1)//2] for idx in xrange(2*nobs) )
        return xsteps, psteps
    def get_steps(self):
        '''Return: 2-tuple of arrays,
        the data values and corresponding cumulative 
        probabilities.
        '''
        xsteps, psteps = self.gen_xp()
        return N.fromiter(xsteps,'f'), N.fromiter(psteps,'f')

Hi Alan,

There is an empiricalcdf function in scipy/sandbox/dhuard/stats.py
It's not fancy but it might do what you want.

David

2007/9/26, Alan Isaac <ai...@am...>:
>
> Is there a standard function or practice for
> plotting the CDF of a series?  (I am aware
> of the output of hist.)
>
> Thank you,
> Alan Isaac
>
>
>
>
>
> -------------------------------------------------------------------------
> This SF.net email is sponsored by: Microsoft
> Defy all challenges. Microsoft(R) Visual Studio 2005.
> http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/
> _______________________________________________
> Matplotlib-users mailing list
> Mat...@li...
> https://lists.sourceforge.net/lists/listinfo/matplotlib-users
>

On Thu, 27 Sep 2007, David Huard apparently wrote:
> scipy/sandbox/dhuard/stats.py 

Well, I also did not think of a double application of 
argsort to rank the observations. Nice.  OK, I've got plenty 
to work with now.

Thanks,
Alan