one more reason to continue working on probability calibration:

https://github.com/scikit-learn/scikit-learn/pull/1176

Alex

On Wed, Apr 10, 2013 at 7:59 PM, Andreas Mueller
<[email protected]> wrote:
> Hi Mathieu.
> From my point of view, producing sensible probabilities is even harder than
> producing
> the correct class labels - so I don't think there is a great solution.
> For binary classification, you could use LogisticRegression, but I'm not
> sure that it really gives
> you good estimates.
>
> Maybe I'd go with the RandomForestClassfier?
>
> Cheers,
> Andy
>
>
> On 04/10/2013 07:21 PM, mathieu lacage wrote:
>
> hi,
>
> I wonder if someone could suggest which classifiers are the most useful to
> produce probabilities associated with their predictions. In my case, I am
> trying to perform a binary classification and I am trying to obtain an
> estimate of the probability that the output classification is correct. I
> initially started from an SVC but it seems that producing probability
> estimates is not one of their biggest strength. I then turned to bayes
> classifiers but the sklearn documentation is pretty clear about the fact
> that the estimates produced by these classifiers are "not good". Is there an
> well known alternative to do this or am I stuck with trying every possible
> option until I find one that appears to work ?
>
> Mathieu
> --
> Mathieu Lacage <[email protected]>
>
>
> ------------------------------------------------------------------------------
> Precog is a next-generation analytics platform capable of advanced
> analytics on semi-structured data. The platform includes APIs for building
> apps and a phenomenal toolset for data science. Developers can use
> our toolset for easy data analysis & visualization. Get a free account!
> http://www2.precog.com/precogplatform/slashdotnewsletter
>
>
>
> _______________________________________________
> Scikit-learn-general mailing list
> [email protected]
> https://lists.sourceforge.net/lists/listinfo/scikit-learn-general
>
>
>
> ------------------------------------------------------------------------------
> Precog is a next-generation analytics platform capable of advanced
> analytics on semi-structured data. The platform includes APIs for building
> apps and a phenomenal toolset for data science. Developers can use
> our toolset for easy data analysis & visualization. Get a free account!
> http://www2.precog.com/precogplatform/slashdotnewsletter
> _______________________________________________
> Scikit-learn-general mailing list
> [email protected]
> https://lists.sourceforge.net/lists/listinfo/scikit-learn-general
>

------------------------------------------------------------------------------
Precog is a next-generation analytics platform capable of advanced
analytics on semi-structured data. The platform includes APIs for building
apps and a phenomenal toolset for data science. Developers can use
our toolset for easy data analysis & visualization. Get a free account!
http://www2.precog.com/precogplatform/slashdotnewsletter
_______________________________________________
Scikit-learn-general mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/scikit-learn-general

Reply via email to