Re: maximum likelihood sensitivity to outliers [message #67121] |
Tue, 07 July 2009 21:24  |
maurya
Messages: 16 Registered: February 2009
|
Junior Member |
|
|
On 8 July, 03:56, Jeremy Bailin <astroco...@gmail.com> wrote:
> This is much more a numerical analysis question than an IDL question,
> but I suspect that some people here can point me in the right
> direction...
>
> I am a big fan of using maximum likelihood estimation to fit
> distributions without binning (and therefore wrote ML_DISTFIT, which
> you can find in JBIU). But one problem is that the method is quite
> sensitive to outliers. Does anyone know (or even better, have an IDL
> implementation of) a more outlier-robust version of the basic maximum
> likelihood estimator?
>
> Thanks,
> -Jeremy.
A maximum-likelihood routine "MLEfit.pro" given at VIRGO web page may
be helpful for you.
http://www.ias.u-psud.fr/virgo/html/IDLpro/MLEfit.pro
Cheers
Maurya
|
|
|
Re: maximum likelihood sensitivity to outliers [message #67218 is a reply to message #67121] |
Wed, 08 July 2009 05:21  |
Jeremy Bailin
Messages: 618 Registered: April 2008
|
Senior Member |
|
|
On Jul 8, 12:24 am, maurya <ramaury...@gmail.com> wrote:
> On 8 July, 03:56, Jeremy Bailin <astroco...@gmail.com> wrote:
>
>> This is much more a numerical analysis question than an IDL question,
>> but I suspect that some people here can point me in the right
>> direction...
>
>> I am a big fan of using maximum likelihood estimation to fit
>> distributions without binning (and therefore wrote ML_DISTFIT, which
>> you can find in JBIU). But one problem is that the method is quite
>> sensitive to outliers. Does anyone know (or even better, have an IDL
>> implementation of) a more outlier-robust version of the basic maximum
>> likelihood estimator?
>
>> Thanks,
>> -Jeremy.
>
> A maximum-likelihood routine "MLEfit.pro" given at VIRGO web page may
> be helpful for you.
>
> http://www.ias.u-psud.fr/virgo/html/IDLpro/MLEfit.pro
>
> Cheers
> Maurya
As far as I can tell, that's a standard ML algorithm, same as mine.
-Jeremy.
|
|
|