how to fit a gamma distribution

[Kushal Kumaran]

On Tue, Apr 14, 2009 at 6:51 AM, ning luwen <ningluwen at gmail.com> wrote:
> i already have tried.
>
> and try to  use the module scipy.optimize.leastsq to fit, but neither
> get a good return or just break out like this.
>
> is there any other good way?or something wrong with my code?
>


I suggest you try the scipy list.  Google for Smart Questions before
you do that, though.

Also, when asking for help on a list, you need to mention what you've
already tried.  You've only wasted a day for a useless answer.

Also, please keep responses on the list.  Other people on the list are
much smarter than I am.

>
>
> Traceback (most recent call last):
>  File "fitgamma.py", line 33, in <lambda>
>    e = lambda v, x, y: (fp(v,x)-y)
>  File "fitgamma.py", line 26, in <lambda>
>    fp = lambda v, x: x**(v[0]-1)/(v[1]**v[0]*gamma(v[0]))*exp(-x/v[1])
>  File "mtrand.pyx", line 1236, in mtrand.RandomState.gamma
> ValueError: shape <= 0
> Traceback (most recent call last):
>  File "fitgamma.py", line 46, in <module>
>    v, success = leastsq(e, v0, args=(xdata,ydata),maxfev=100000)
>  File "/usr/lib/python2.6/dist-packages/scipy/optimize/minpack.py",
> line 270, in leastsq
>    retval = _minpack._lmdif(func,x0,args,full_output,ftol,xtol,gtol,maxfev,epsfcn,factor,diag)
> minpack.error: Error occured while calling the Python function named <lambda>
>
>
>
> ps:the part code in fitting the distribution(xdata,ydata are vectors
> of the data need to fit.
>
> fp = lambda v, x: x**(v[0]-1)/(v[1]**v[0]*gamma(v[0]))*exp(-x/v[1])
> e = lambda v, x, y: (fp(v,x)-y)
> v0 = [6, 3]
> v, success = leastsq(e, v0, args=(xdata,ydata),maxfev=100000)
>
>
>
>
>
> On Mon, Apr 13, 2009 at 8:49 PM, Kushal Kumaran
> <kushal.kumaran at gmail.com> wrote:
>> On Mon, Apr 13, 2009 at 1:40 PM, ning luwen <ningluwen at gmail.com> wrote:
>>> hi,
>>>  i need to fit a gamma distribution, is there any module can do the job?
>>>
>>
>> Did you try google?  Searching for python curve fitting returns some
>> promising results.
>>
>> --
>> kushal
>> --
>> http://mail.python.org/mailman/listinfo/python-list
>>
>