From andreasmuellerml at gmail.com  Wed Dec  2 19:35:45 2020
From: andreasmuellerml at gmail.com (Andreas C. Mueller)
Date: Thu, 03 Dec 2020 00:35:45 +0000
Subject: [scikit-learn] Changes in Travis billing
In-Reply-To: <20201126141212.x72eecggoozdocck@phare.normalesup.org>
References: <20201102105018.kfarylwpao6iliju@phare.normalesup.org>
 <CAEOrW49QWQLKi0Rd5AdESD=YCwsm356egRdH+=fxtqZeiMZNzw@mail.gmail.com>
 <CAFvE7K6dPVtH4e6=0toWujbLiByg=mVOxayh3rTvEUNZCt8q+g@mail.gmail.com>
 <CAEOrW49SNzEffyWNOLTahokJiGQoFCM7bx+s5GYjMrNnw2ZrFg@mail.gmail.com>
 <20201126140652.tdrsjzfasb64kqua@phare.normalesup.org>
 <20201126141212.x72eecggoozdocck@phare.normalesup.org>
Message-ID: <emdf0ee827-9920-429c-ad0a-d15556e8557a@desktop-fo3usmm>

Sorry I'm probably missing some detail but what does travis provide that 
github actions and azure pipeline don't provide?

------ Original Message ------
From: "Gael Varoquaux" <gael.varoquaux at normalesup.org>
To: "Scikit-learn mailing list" <scikit-learn at python.org>
Sent: 11/26/2020 6:12:12 AM
Subject: Re: [scikit-learn] Changes in Travis billing

>On Thu, Nov 26, 2020 at 03:06:52PM +0100, Gael Varoquaux wrote:
>>  On Thu, Nov 26, 2020 at 02:45:33PM +0100, Adrin wrote:
>>  > At this point I'm at a loss, and reading the NumFocus chat and other
>>  > packages' experience with them on the same topic, seems like we just
>>  > need to move out of Travis.
>
>>  Agreed. Do we still need them for something essential?
>
>Sorry, ARM, it was just above in the conversation.
>
>I think that we have no other option than reduce the frequency of the
>cron, and wait for other platforms to offer ARM, which will hopefully
>happen soonish.
>
>G
>_______________________________________________
>scikit-learn mailing list
>scikit-learn at python.org
>https://mail.python.org/mailman/listinfo/scikit-learn


From g.lemaitre58 at gmail.com  Thu Dec  3 04:15:50 2020
From: g.lemaitre58 at gmail.com (=?UTF-8?Q?Guillaume_Lema=C3=AEtre?=)
Date: Thu, 3 Dec 2020 10:15:50 +0100
Subject: [scikit-learn] Changes in Travis billing
In-Reply-To: <emdf0ee827-9920-429c-ad0a-d15556e8557a@desktop-fo3usmm>
References: <20201102105018.kfarylwpao6iliju@phare.normalesup.org>
 <CAEOrW49QWQLKi0Rd5AdESD=YCwsm356egRdH+=fxtqZeiMZNzw@mail.gmail.com>
 <CAFvE7K6dPVtH4e6=0toWujbLiByg=mVOxayh3rTvEUNZCt8q+g@mail.gmail.com>
 <CAEOrW49SNzEffyWNOLTahokJiGQoFCM7bx+s5GYjMrNnw2ZrFg@mail.gmail.com>
 <20201126140652.tdrsjzfasb64kqua@phare.normalesup.org>
 <20201126141212.x72eecggoozdocck@phare.normalesup.org>
 <emdf0ee827-9920-429c-ad0a-d15556e8557a@desktop-fo3usmm>
Message-ID: <CACDxx9gSGfoKOQMds2pkmSKVFR2eAkP3EVmxrLWPrwu4u5hwMg@mail.gmail.com>

ARM support

On Thu, 3 Dec 2020 at 01:37, Andreas C. Mueller <andreasmuellerml at gmail.com>
wrote:

> Sorry I'm probably missing some detail but what does travis provide that
> github actions and azure pipeline don't provide?
>
> ------ Original Message ------
> From: "Gael Varoquaux" <gael.varoquaux at normalesup.org>
> To: "Scikit-learn mailing list" <scikit-learn at python.org>
> Sent: 11/26/2020 6:12:12 AM
> Subject: Re: [scikit-learn] Changes in Travis billing
>
> >On Thu, Nov 26, 2020 at 03:06:52PM +0100, Gael Varoquaux wrote:
> >>  On Thu, Nov 26, 2020 at 02:45:33PM +0100, Adrin wrote:
> >>  > At this point I'm at a loss, and reading the NumFocus chat and other
> >>  > packages' experience with them on the same topic, seems like we just
> >>  > need to move out of Travis.
> >
> >>  Agreed. Do we still need them for something essential?
> >
> >Sorry, ARM, it was just above in the conversation.
> >
> >I think that we have no other option than reduce the frequency of the
> >cron, and wait for other platforms to offer ARM, which will hopefully
> >happen soonish.
> >
> >G
> >_______________________________________________
> >scikit-learn mailing list
> >scikit-learn at python.org
> >https://mail.python.org/mailman/listinfo/scikit-learn
>
> _______________________________________________
> scikit-learn mailing list
> scikit-learn at python.org
> https://mail.python.org/mailman/listinfo/scikit-learn
>


-- 
Guillaume Lemaitre
Scikit-learn @ Inria Foundation
https://glemaitre.github.io/
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mail.python.org/pipermail/scikit-learn/attachments/20201203/be584358/attachment.html>

From adrin.jalali at gmail.com  Thu Dec  3 05:36:32 2020
From: adrin.jalali at gmail.com (Adrin)
Date: Thu, 3 Dec 2020 11:36:32 +0100
Subject: [scikit-learn] Changes in Travis billing
In-Reply-To: <CACDxx9gSGfoKOQMds2pkmSKVFR2eAkP3EVmxrLWPrwu4u5hwMg@mail.gmail.com>
References: <20201102105018.kfarylwpao6iliju@phare.normalesup.org>
 <CAEOrW49QWQLKi0Rd5AdESD=YCwsm356egRdH+=fxtqZeiMZNzw@mail.gmail.com>
 <CAFvE7K6dPVtH4e6=0toWujbLiByg=mVOxayh3rTvEUNZCt8q+g@mail.gmail.com>
 <CAEOrW49SNzEffyWNOLTahokJiGQoFCM7bx+s5GYjMrNnw2ZrFg@mail.gmail.com>
 <20201126140652.tdrsjzfasb64kqua@phare.normalesup.org>
 <20201126141212.x72eecggoozdocck@phare.normalesup.org>
 <emdf0ee827-9920-429c-ad0a-d15556e8557a@desktop-fo3usmm>
 <CACDxx9gSGfoKOQMds2pkmSKVFR2eAkP3EVmxrLWPrwu4u5hwMg@mail.gmail.com>
Message-ID: <CAEOrW4-437koMoQ=Y-uT4W7Cp74Wg4n9Qktm=CM3f02142vXuQ@mail.gmail.com>

I got a response from Travis, and according to this, we don't qualify for
the free credits:

The free plan will grant your organization 10000 credits.

We offer an Open Source Subscription for free to non-commercial open-source
projects. To qualify for an Open Source subscription, the project must meet
the following requirements:


   - You are a project lead or regular committer (latest commit in the last
   month)
   - Project must be at least 3 months old and is in active development
   (with regular commits and activity)
   - Project meets the OSD <https://opensource.org/docs/osd> specification
   -
*Project must not be sponsored by a commercial company or organization
   (monetary or with employees paid to work on the project) *
   - Project can not provide commercial services or distribute paid
   versions of the software


Does this sound like you and your project? We'd be very happy to support
you!

However, if your project does not match these requirements or you have
further questions [1], please feel free to ask!

We look forward to your response if you meet these requirements to proceed
with the next steps.

Thank you

On Thu, Dec 3, 2020 at 10:17 AM Guillaume Lema?tre <g.lemaitre58 at gmail.com>
wrote:

> ARM support
>
> On Thu, 3 Dec 2020 at 01:37, Andreas C. Mueller <
> andreasmuellerml at gmail.com> wrote:
>
>> Sorry I'm probably missing some detail but what does travis provide that
>> github actions and azure pipeline don't provide?
>>
>> ------ Original Message ------
>> From: "Gael Varoquaux" <gael.varoquaux at normalesup.org>
>> To: "Scikit-learn mailing list" <scikit-learn at python.org>
>> Sent: 11/26/2020 6:12:12 AM
>> Subject: Re: [scikit-learn] Changes in Travis billing
>>
>> >On Thu, Nov 26, 2020 at 03:06:52PM +0100, Gael Varoquaux wrote:
>> >>  On Thu, Nov 26, 2020 at 02:45:33PM +0100, Adrin wrote:
>> >>  > At this point I'm at a loss, and reading the NumFocus chat and other
>> >>  > packages' experience with them on the same topic, seems like we just
>> >>  > need to move out of Travis.
>> >
>> >>  Agreed. Do we still need them for something essential?
>> >
>> >Sorry, ARM, it was just above in the conversation.
>> >
>> >I think that we have no other option than reduce the frequency of the
>> >cron, and wait for other platforms to offer ARM, which will hopefully
>> >happen soonish.
>> >
>> >G
>> >_______________________________________________
>> >scikit-learn mailing list
>> >scikit-learn at python.org
>> >https://mail.python.org/mailman/listinfo/scikit-learn
>>
>> _______________________________________________
>> scikit-learn mailing list
>> scikit-learn at python.org
>> https://mail.python.org/mailman/listinfo/scikit-learn
>>
>
>
> --
> Guillaume Lemaitre
> Scikit-learn @ Inria Foundation
> https://glemaitre.github.io/
> _______________________________________________
> scikit-learn mailing list
> scikit-learn at python.org
> https://mail.python.org/mailman/listinfo/scikit-learn
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mail.python.org/pipermail/scikit-learn/attachments/20201203/bffd8019/attachment-0001.html>

From nelle.varoquaux at gmail.com  Thu Dec  3 05:47:39 2020
From: nelle.varoquaux at gmail.com (Nelle Varoquaux)
Date: Thu, 3 Dec 2020 11:47:39 +0100
Subject: [scikit-learn] Changes in Travis billing
In-Reply-To: <CAEOrW4-437koMoQ=Y-uT4W7Cp74Wg4n9Qktm=CM3f02142vXuQ@mail.gmail.com>
References: <20201102105018.kfarylwpao6iliju@phare.normalesup.org>
 <CAEOrW49QWQLKi0Rd5AdESD=YCwsm356egRdH+=fxtqZeiMZNzw@mail.gmail.com>
 <CAFvE7K6dPVtH4e6=0toWujbLiByg=mVOxayh3rTvEUNZCt8q+g@mail.gmail.com>
 <CAEOrW49SNzEffyWNOLTahokJiGQoFCM7bx+s5GYjMrNnw2ZrFg@mail.gmail.com>
 <20201126140652.tdrsjzfasb64kqua@phare.normalesup.org>
 <20201126141212.x72eecggoozdocck@phare.normalesup.org>
 <emdf0ee827-9920-429c-ad0a-d15556e8557a@desktop-fo3usmm>
 <CACDxx9gSGfoKOQMds2pkmSKVFR2eAkP3EVmxrLWPrwu4u5hwMg@mail.gmail.com>
 <CAEOrW4-437koMoQ=Y-uT4W7Cp74Wg4n9Qktm=CM3f02142vXuQ@mail.gmail.com>
Message-ID: <CAE-UAvTdYFZE+rvQTVFtCAmoafJea-dHsf12m+jf+1duAVkvxg@mail.gmail.com>

Wow? what?! That's insane?
scikit-image got some credits (
https://github.com/scikit-image/scikit-image-wheels/pull/47#issuecomment-736760539)
without much issue? Maybe someone should reach out directly on this thread
to the travis people?

Cheers,
N

On Thu, 3 Dec 2020 at 11:37, Adrin <adrin.jalali at gmail.com> wrote:

> I got a response from Travis, and according to this, we don't qualify for
> the free credits:
>
> The free plan will grant your organization 10000 credits.
>
> We offer an Open Source Subscription for free to non-commercial
> open-source projects. To qualify for an Open Source subscription, the
> project must meet the following requirements:
>
>
>    - You are a project lead or regular committer (latest commit in the
>    last month)
>    - Project must be at least 3 months old and is in active development
>    (with regular commits and activity)
>    - Project meets the OSD <https://opensource.org/docs/osd> specification
>    -
> *Project must not be sponsored by a commercial company or organization
>    (monetary or with employees paid to work on the project) *
>    - Project can not provide commercial services or distribute paid
>    versions of the software
>
>
> Does this sound like you and your project? We'd be very happy to support
> you!
>
> However, if your project does not match these requirements or you have
> further questions [1], please feel free to ask!
>
> We look forward to your response if you meet these requirements to proceed
> with the next steps.
>
> Thank you
>
> On Thu, Dec 3, 2020 at 10:17 AM Guillaume Lema?tre <g.lemaitre58 at gmail.com>
> wrote:
>
>> ARM support
>>
>> On Thu, 3 Dec 2020 at 01:37, Andreas C. Mueller <
>> andreasmuellerml at gmail.com> wrote:
>>
>>> Sorry I'm probably missing some detail but what does travis provide that
>>> github actions and azure pipeline don't provide?
>>>
>>> ------ Original Message ------
>>> From: "Gael Varoquaux" <gael.varoquaux at normalesup.org>
>>> To: "Scikit-learn mailing list" <scikit-learn at python.org>
>>> Sent: 11/26/2020 6:12:12 AM
>>> Subject: Re: [scikit-learn] Changes in Travis billing
>>>
>>> >On Thu, Nov 26, 2020 at 03:06:52PM +0100, Gael Varoquaux wrote:
>>> >>  On Thu, Nov 26, 2020 at 02:45:33PM +0100, Adrin wrote:
>>> >>  > At this point I'm at a loss, and reading the NumFocus chat and
>>> other
>>> >>  > packages' experience with them on the same topic, seems like we
>>> just
>>> >>  > need to move out of Travis.
>>> >
>>> >>  Agreed. Do we still need them for something essential?
>>> >
>>> >Sorry, ARM, it was just above in the conversation.
>>> >
>>> >I think that we have no other option than reduce the frequency of the
>>> >cron, and wait for other platforms to offer ARM, which will hopefully
>>> >happen soonish.
>>> >
>>> >G
>>> >_______________________________________________
>>> >scikit-learn mailing list
>>> >scikit-learn at python.org
>>> >https://mail.python.org/mailman/listinfo/scikit-learn
>>>
>>> _______________________________________________
>>> scikit-learn mailing list
>>> scikit-learn at python.org
>>> https://mail.python.org/mailman/listinfo/scikit-learn
>>>
>>
>>
>> --
>> Guillaume Lemaitre
>> Scikit-learn @ Inria Foundation
>> https://glemaitre.github.io/
>> _______________________________________________
>> scikit-learn mailing list
>> scikit-learn at python.org
>> https://mail.python.org/mailman/listinfo/scikit-learn
>>
> _______________________________________________
> scikit-learn mailing list
> scikit-learn at python.org
> https://mail.python.org/mailman/listinfo/scikit-learn
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mail.python.org/pipermail/scikit-learn/attachments/20201203/d26c9695/attachment.html>

From solegalli at protonmail.com  Thu Dec  3 05:55:38 2020
From: solegalli at protonmail.com (Sole Galli)
Date: Thu, 03 Dec 2020 10:55:38 +0000
Subject: [scikit-learn] sample_weight vs class_weight
Message-ID: <gTEPm7FOtadRFt4Uh95vOMgl-21MkzmVmHKmpVMYV0g6GGuEWs7PoK355BXAFEFg8Fq6gRNbxJFJQxe3kyCqQDl60JdYuQ3pOxboYSHGwfI=@protonmail.com>

Hello team,

What is the difference in the implementation of class_weight and sample_weight in those algorithms that support both? like random forest or logistic regression?

Are both modifying the loss function? in a similar way?

Thank you!

Sole
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mail.python.org/pipermail/scikit-learn/attachments/20201203/7d01266e/attachment.html>

From olivier.grisel at ensta.org  Thu Dec  3 13:20:52 2020
From: olivier.grisel at ensta.org (Olivier Grisel)
Date: Thu, 3 Dec 2020 19:20:52 +0100
Subject: [scikit-learn] [ANN] scikit-learn 0.24.0rc1 is online!
Message-ID: <CAFvE7K7XJShY9=Y_G-_kim7S-4DkWd4iVjF-n3uCwFUtYR6Oew@mail.gmail.com>

Please help us test the first release candidate for scikit-learn 0.24.0:

   pip install scikit-learn==0.24.0rc1

Changelog: https://scikit-learn.org/0.24/whats_new/v0.24.html

In particular, if you maintain a project with a dependency on
scikit-learn, please let us know about any regression.

Feel free to also retweet the announcement to get more people to test
it before the final release (probably in 1 week or 2):

  https://twitter.com/scikit_learn/status/1334562221498753026

Thanks to anybody who helped make this happen!

-- 
Olivier

From bsipocz at gmail.com  Thu Dec  3 14:59:32 2020
From: bsipocz at gmail.com (Brigitta Sipocz)
Date: Thu, 3 Dec 2020 11:59:32 -0800
Subject: [scikit-learn] Changes in Travis billing
In-Reply-To: <CAE-UAvTdYFZE+rvQTVFtCAmoafJea-dHsf12m+jf+1duAVkvxg@mail.gmail.com>
References: <20201102105018.kfarylwpao6iliju@phare.normalesup.org>
 <CAEOrW49QWQLKi0Rd5AdESD=YCwsm356egRdH+=fxtqZeiMZNzw@mail.gmail.com>
 <CAFvE7K6dPVtH4e6=0toWujbLiByg=mVOxayh3rTvEUNZCt8q+g@mail.gmail.com>
 <CAEOrW49SNzEffyWNOLTahokJiGQoFCM7bx+s5GYjMrNnw2ZrFg@mail.gmail.com>
 <20201126140652.tdrsjzfasb64kqua@phare.normalesup.org>
 <20201126141212.x72eecggoozdocck@phare.normalesup.org>
 <emdf0ee827-9920-429c-ad0a-d15556e8557a@desktop-fo3usmm>
 <CACDxx9gSGfoKOQMds2pkmSKVFR2eAkP3EVmxrLWPrwu4u5hwMg@mail.gmail.com>
 <CAEOrW4-437koMoQ=Y-uT4W7Cp74Wg4n9Qktm=CM3f02142vXuQ@mail.gmail.com>
 <CAE-UAvTdYFZE+rvQTVFtCAmoafJea-dHsf12m+jf+1duAVkvxg@mail.gmail.com>
Message-ID: <CAEjJ3fhMtFe3JaQnusA12xFTE=Y3mGdBr+U=3vs3nxwnoaHWtA@mail.gmail.com>

Hi,

Astropy also got a (one time) credit, that we very quickly burned up. And
NumPy also got a credit, but I'm not sure whether it's also a one-off or a
monthly.

For astropy, our workaround is to run the tests on aarch64 and the other
more exotic hardware from a weekly cron, as each takes a very long time on
GH Actions. Hopefully, a better, native solution will come to Actions/Azure
soon enough.
I link the PR that set this up, maybe this solution would be good enough
for sklearn, too.

https://github.com/astropy/astropy/pull/11045


Cheers,
 Brigitta

On Thu, 3 Dec 2020 at 02:50, Nelle Varoquaux <nelle.varoquaux at gmail.com>
wrote:

> Wow? what?! That's insane?
> scikit-image got some credits (
> https://github.com/scikit-image/scikit-image-wheels/pull/47#issuecomment-736760539)
> without much issue? Maybe someone should reach out directly on this thread
> to the travis people?
>
> Cheers,
> N
>
> On Thu, 3 Dec 2020 at 11:37, Adrin <adrin.jalali at gmail.com> wrote:
>
>> I got a response from Travis, and according to this, we don't qualify for
>> the free credits:
>>
>> The free plan will grant your organization 10000 credits.
>>
>> We offer an Open Source Subscription for free to non-commercial
>> open-source projects. To qualify for an Open Source subscription, the
>> project must meet the following requirements:
>>
>>
>>    - You are a project lead or regular committer (latest commit in the
>>    last month)
>>    - Project must be at least 3 months old and is in active development
>>    (with regular commits and activity)
>>    - Project meets the OSD <https://opensource.org/docs/osd>
>>    specification
>>    -
>> *Project must not be sponsored by a commercial company or organization
>>    (monetary or with employees paid to work on the project) *
>>    - Project can not provide commercial services or distribute paid
>>    versions of the software
>>
>>
>> Does this sound like you and your project? We'd be very happy to support
>> you!
>>
>> However, if your project does not match these requirements or you have
>> further questions [1], please feel free to ask!
>>
>> We look forward to your response if you meet these requirements to
>> proceed with the next steps.
>>
>> Thank you
>>
>> On Thu, Dec 3, 2020 at 10:17 AM Guillaume Lema?tre <
>> g.lemaitre58 at gmail.com> wrote:
>>
>>> ARM support
>>>
>>> On Thu, 3 Dec 2020 at 01:37, Andreas C. Mueller <
>>> andreasmuellerml at gmail.com> wrote:
>>>
>>>> Sorry I'm probably missing some detail but what does travis provide
>>>> that
>>>> github actions and azure pipeline don't provide?
>>>>
>>>> ------ Original Message ------
>>>> From: "Gael Varoquaux" <gael.varoquaux at normalesup.org>
>>>> To: "Scikit-learn mailing list" <scikit-learn at python.org>
>>>> Sent: 11/26/2020 6:12:12 AM
>>>> Subject: Re: [scikit-learn] Changes in Travis billing
>>>>
>>>> >On Thu, Nov 26, 2020 at 03:06:52PM +0100, Gael Varoquaux wrote:
>>>> >>  On Thu, Nov 26, 2020 at 02:45:33PM +0100, Adrin wrote:
>>>> >>  > At this point I'm at a loss, and reading the NumFocus chat and
>>>> other
>>>> >>  > packages' experience with them on the same topic, seems like we
>>>> just
>>>> >>  > need to move out of Travis.
>>>> >
>>>> >>  Agreed. Do we still need them for something essential?
>>>> >
>>>> >Sorry, ARM, it was just above in the conversation.
>>>> >
>>>> >I think that we have no other option than reduce the frequency of the
>>>> >cron, and wait for other platforms to offer ARM, which will hopefully
>>>> >happen soonish.
>>>> >
>>>> >G
>>>> >_______________________________________________
>>>> >scikit-learn mailing list
>>>> >scikit-learn at python.org
>>>> >https://mail.python.org/mailman/listinfo/scikit-learn
>>>>
>>>> _______________________________________________
>>>> scikit-learn mailing list
>>>> scikit-learn at python.org
>>>> https://mail.python.org/mailman/listinfo/scikit-learn
>>>>
>>>
>>>
>>> --
>>> Guillaume Lemaitre
>>> Scikit-learn @ Inria Foundation
>>> https://glemaitre.github.io/
>>> _______________________________________________
>>> scikit-learn mailing list
>>> scikit-learn at python.org
>>> https://mail.python.org/mailman/listinfo/scikit-learn
>>>
>> _______________________________________________
>> scikit-learn mailing list
>> scikit-learn at python.org
>> https://mail.python.org/mailman/listinfo/scikit-learn
>>
> _______________________________________________
> scikit-learn mailing list
> scikit-learn at python.org
> https://mail.python.org/mailman/listinfo/scikit-learn
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mail.python.org/pipermail/scikit-learn/attachments/20201203/914b9e3e/attachment.html>

From solegalli at protonmail.com  Fri Dec  4 05:40:44 2020
From: solegalli at protonmail.com (Sole Galli)
Date: Fri, 04 Dec 2020 10:40:44 +0000
Subject: [scikit-learn] sample_weight vs class_weight
In-Reply-To: <gTEPm7FOtadRFt4Uh95vOMgl-21MkzmVmHKmpVMYV0g6GGuEWs7PoK355BXAFEFg8Fq6gRNbxJFJQxe3kyCqQDl60JdYuQ3pOxboYSHGwfI=@protonmail.com>
References: <gTEPm7FOtadRFt4Uh95vOMgl-21MkzmVmHKmpVMYV0g6GGuEWs7PoK355BXAFEFg8Fq6gRNbxJFJQxe3kyCqQDl60JdYuQ3pOxboYSHGwfI=@protonmail.com>
Message-ID: <xHCo3nlA5pIjSkEFpDmhZXy0YnWvLhNH9Ev2V-yuhfBVAFyr-KEwcGeztAaR3MRWq1jM2VNOcNRRHBB5afeFzPwVz522sp7SNUHsTRSYhzc=@protonmail.com>

Actually, I found the answer. Both seem to be optimising the loss function for the various algorithms, below I include some links.

If, we pass class_weight and sample_weight, then the final cost / weight is a combination of both.

I have a follow up question: in which scenario would we use both? why do some estimators allow to pass weights both as a dict in the init or as sample weights in fit? what's the logic? I found it a bit confusing at the beginning.

Thank you!

https://stackoverflow.com/questions/30805192/scikit-learn-random-forest-class-weight-and-sample-weight-parameters

https://stackoverflow.com/questions/30972029/how-does-the-class-weight-parameter-in-scikit-learn-work/30982811#30982811

Soledad Galli
https://www.trainindata.com/

??????? Original Message ???????
On Thursday, December 3, 2020 11:55 AM, Sole Galli via scikit-learn <scikit-learn at python.org> wrote:

> Hello team,
>
> What is the difference in the implementation of class_weight and sample_weight in those algorithms that support both? like random forest or logistic regression?
>
> Are both modifying the loss function? in a similar way?
>
> Thank you!
>
> Sole
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mail.python.org/pipermail/scikit-learn/attachments/20201204/50e98e28/attachment-0001.html>

From niourf at gmail.com  Fri Dec  4 05:59:03 2020
From: niourf at gmail.com (Nicolas Hug)
Date: Fri, 4 Dec 2020 10:59:03 +0000
Subject: [scikit-learn] sample_weight vs class_weight
In-Reply-To: <xHCo3nlA5pIjSkEFpDmhZXy0YnWvLhNH9Ev2V-yuhfBVAFyr-KEwcGeztAaR3MRWq1jM2VNOcNRRHBB5afeFzPwVz522sp7SNUHsTRSYhzc=@protonmail.com>
References: <gTEPm7FOtadRFt4Uh95vOMgl-21MkzmVmHKmpVMYV0g6GGuEWs7PoK355BXAFEFg8Fq6gRNbxJFJQxe3kyCqQDl60JdYuQ3pOxboYSHGwfI=@protonmail.com>
 <xHCo3nlA5pIjSkEFpDmhZXy0YnWvLhNH9Ev2V-yuhfBVAFyr-KEwcGeztAaR3MRWq1jM2VNOcNRRHBB5afeFzPwVz522sp7SNUHsTRSYhzc=@protonmail.com>
Message-ID: <bfb529be-6de4-f6dc-ec68-821bade7ce25@gmail.com>

Basically passing class weights should be equivalent to passing 
per-class-constant sample weights.

 > why do some estimators allow to pass weights both as a dict in the 
init or as sample weights in fit? what's the logic?

SW is a per-sample property (aligned with X and y) so we avoid passing 
those to init because the data isn't known when initializing the 
estimator. It's only known when calling fit. In general we avoid passing 
data-related info into init so that the same instance can be fitted on 
any data (with different number of samples, different classes, etc.).

We allow to pass class_weight in init because the 'balanced' option is 
data-agnostic. Arguably, allowing a dict with actual class values 
violates the above argument (of not having data-related stuff in init), 
so I guess that's where the logic ends ;)

As to why one would use both, I'm not so sure honestly.

Nicolas


On 12/4/20 10:40 AM, Sole Galli via scikit-learn wrote:
> Actually, I found the answer. Both seem to be optimising the loss 
> function for the various algorithms, below I include some links.
>
> If, we pass *class_weight* and *sample_weight,* then the final cost / 
> weight is a combination of both.
>
> I have a follow up question: in which scenario would we use both? why 
> do some estimators allow to pass weights both as a dict in the init or 
> as sample weights in fit? what's the logic? I found it a bit confusing 
> at the beginning.
>
> Thank you!
>
> https://stackoverflow.com/questions/30805192/scikit-learn-random-forest-class-weight-and-sample-weight-parameters
>
> https://stackoverflow.com/questions/30972029/how-does-the-class-weight-parameter-in-scikit-learn-work/30982811#30982811
>
> Soledad Galli
> https://www.trainindata.com/
>
>
> ??????? Original Message ???????
> On Thursday, December 3, 2020 11:55 AM, Sole Galli via scikit-learn 
> <scikit-learn at python.org> wrote:
>
>> Hello team,
>>
>> What is the difference in the implementation of class_weight and 
>> sample_weight in those algorithms that support both? like random 
>> forest or logistic regression?
>>
>> Are both modifying the loss function? in a similar way?
>>
>> Thank you!
>>
>> Sole
>>
>>
>
>
> _______________________________________________
> scikit-learn mailing list
> scikit-learn at python.org
> https://mail.python.org/mailman/listinfo/scikit-learn
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mail.python.org/pipermail/scikit-learn/attachments/20201204/66684009/attachment.html>

From maykonschots at gmail.com  Fri Dec  4 06:06:54 2020
From: maykonschots at gmail.com (mrschots)
Date: Fri, 4 Dec 2020 12:06:54 +0100
Subject: [scikit-learn] sample_weight vs class_weight
In-Reply-To: <bfb529be-6de4-f6dc-ec68-821bade7ce25@gmail.com>
References: <gTEPm7FOtadRFt4Uh95vOMgl-21MkzmVmHKmpVMYV0g6GGuEWs7PoK355BXAFEFg8Fq6gRNbxJFJQxe3kyCqQDl60JdYuQ3pOxboYSHGwfI=@protonmail.com>
 <xHCo3nlA5pIjSkEFpDmhZXy0YnWvLhNH9Ev2V-yuhfBVAFyr-KEwcGeztAaR3MRWq1jM2VNOcNRRHBB5afeFzPwVz522sp7SNUHsTRSYhzc=@protonmail.com>
 <bfb529be-6de4-f6dc-ec68-821bade7ce25@gmail.com>
Message-ID: <CANThK3+3u3r5kJBCmJLs=k=_o00+YdKszqGY-yFCsd=-yRJ_oA@mail.gmail.com>

I have been using both in time-series classification. I put a exponential
decay in sample_weights AND class weights as a dictionary.

BR/Schots

Em sex., 4 de dez. de 2020 ?s 12:01, Nicolas Hug <niourf at gmail.com>
escreveu:

> Basically passing class weights should be equivalent to passing
> per-class-constant sample weights.
>
> > why do some estimators allow to pass weights both as a dict in the init
> or as sample weights in fit? what's the logic?
>
> SW is a per-sample property (aligned with X and y) so we avoid passing
> those to init because the data isn't known when initializing the estimator.
> It's only known when calling fit. In general we avoid passing data-related
> info into init so that the same instance can be fitted on any data (with
> different number of samples, different classes, etc.).
>
> We allow to pass class_weight in init because the 'balanced' option is
> data-agnostic. Arguably, allowing a dict with actual class values violates
> the above argument (of not having data-related stuff in init), so I guess
> that's where the logic ends ;)
>
> As to why one would use both, I'm not so sure honestly.
>
>
> Nicolas
>
>
> On 12/4/20 10:40 AM, Sole Galli via scikit-learn wrote:
>
> Actually, I found the answer. Both seem to be optimising the loss function
> for the various algorithms, below I include some links.
>
> If, we pass *class_weight* and *sample_weight,* then the final cost /
> weight is a combination of both.
>
> I have a follow up question: in which scenario would we use both? why do
> some estimators allow to pass weights both as a dict in the init or as
> sample weights in fit? what's the logic? I found it a bit confusing at the
> beginning.
>
> Thank you!
>
>
> https://stackoverflow.com/questions/30805192/scikit-learn-random-forest-class-weight-and-sample-weight-parameters
>
>
> https://stackoverflow.com/questions/30972029/how-does-the-class-weight-parameter-in-scikit-learn-work/30982811#30982811
>
> Soledad Galli
> https://www.trainindata.com/
>
>
> ??????? Original Message ???????
> On Thursday, December 3, 2020 11:55 AM, Sole Galli via scikit-learn
> <scikit-learn at python.org> <scikit-learn at python.org> wrote:
>
> Hello team,
>
> What is the difference in the implementation of class_weight and
> sample_weight in those algorithms that support both? like random forest or
> logistic regression?
>
> Are both modifying the loss function? in a similar way?
>
> Thank you!
>
> Sole
>
>
>
>
> _______________________________________________
> scikit-learn mailing listscikit-learn at python.orghttps://mail.python.org/mailman/listinfo/scikit-learn
>
> _______________________________________________
> scikit-learn mailing list
> scikit-learn at python.org
> https://mail.python.org/mailman/listinfo/scikit-learn
>
-- 
Schots
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mail.python.org/pipermail/scikit-learn/attachments/20201204/95ae467d/attachment.html>

From gael.varoquaux at normalesup.org  Fri Dec  4 16:06:36 2020
From: gael.varoquaux at normalesup.org (Gael Varoquaux)
Date: Fri, 4 Dec 2020 22:06:36 +0100
Subject: [scikit-learn] Presented scikit-learn to the French President
Message-ID: <20201204210636.36geowufvvphujmg@phare.normalesup.org>

Hi scikit-learn community,

Today, I presented some efforts in digital health to the French president
and part of the government. As these efforts were partly powered by
scikit-learn (and the whole pydata stack, to be fair), the team in charge
of the event had printed a huge scikit-learn logo behind me:
https://twitter.com/GaelVaroquaux/status/1334959438059462659 (terrible
mobile-phone picture)

I would have liked to get a picture with the president and the logo, but
it seems that they are releasing only a handful of pictures :(. Anyhow... 


Thanks to the community! This is a huge success. For health topics (we
are talking nationwide electronic health records) the ability to build on
an independent open-source stack is extremely important. We, as a wider
community, are building something priceless.

Cheers,

Ga?l

From adrin.jalali at gmail.com  Fri Dec  4 18:08:20 2020
From: adrin.jalali at gmail.com (Adrin)
Date: Sat, 5 Dec 2020 00:08:20 +0100
Subject: [scikit-learn] Presented scikit-learn to the French President
In-Reply-To: <20201204210636.36geowufvvphujmg@phare.normalesup.org>
References: <20201204210636.36geowufvvphujmg@phare.normalesup.org>
Message-ID: <CAEOrW49m_HXnNShBOHOA8EuiPXtBwQPaOF9f32gVqB7=aB1iUw@mail.gmail.com>

Nice, such a milestone!

On Fri., Dec. 4, 2020, 23:59 Gael Varoquaux, <gael.varoquaux at normalesup.org>
wrote:

> Hi scikit-learn community,
>
> Today, I presented some efforts in digital health to the French president
> and part of the government. As these efforts were partly powered by
> scikit-learn (and the whole pydata stack, to be fair), the team in charge
> of the event had printed a huge scikit-learn logo behind me:
> https://twitter.com/GaelVaroquaux/status/1334959438059462659 (terrible
> mobile-phone picture)
>
> I would have liked to get a picture with the president and the logo, but
> it seems that they are releasing only a handful of pictures :(. Anyhow...
>
>
> Thanks to the community! This is a huge success. For health topics (we
> are talking nationwide electronic health records) the ability to build on
> an independent open-source stack is extremely important. We, as a wider
> community, are building something priceless.
>
> Cheers,
>
> Ga?l
> _______________________________________________
> scikit-learn mailing list
> scikit-learn at python.org
> https://mail.python.org/mailman/listinfo/scikit-learn
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mail.python.org/pipermail/scikit-learn/attachments/20201205/cf526882/attachment.html>

From olivertomic at zoho.com  Fri Dec  4 18:22:50 2020
From: olivertomic at zoho.com (Oliver Tomic)
Date: Sat, 05 Dec 2020 00:22:50 +0100
Subject: [scikit-learn] Presented scikit-learn to the French President
In-Reply-To: <CAEOrW49m_HXnNShBOHOA8EuiPXtBwQPaOF9f32gVqB7=aB1iUw@mail.gmail.com>
References: <20201204210636.36geowufvvphujmg@phare.normalesup.org>
 <CAEOrW49m_HXnNShBOHOA8EuiPXtBwQPaOF9f32gVqB7=aB1iUw@mail.gmail.com>
Message-ID: <1763010b5df.1049ed1a417502.9100953640053855241@zoho.com>

That's brilliant! Good to know that scikit-learn is increasingly visible outside the ML community and that even the French president heard about it now.?



Thanks a lot to the community for this fantastic package. We use scikit-learn extensively in our ML course at the Norwegian University of Life Sciences and our students love it.?



Cheers

Oliver









---- On Sat, 05 Dec 2020 00:08:20 +0100 Adrin <adrin.jalali at gmail.com> wrote ----



Nice, such a milestone!



On Fri., Dec. 4, 2020, 23:59 Gael Varoquaux, <mailto:gael.varoquaux at normalesup.org> wrote:

Hi scikit-learn community,
 
 Today, I presented some efforts in digital health to the French president
 and part of the government. As these efforts were partly powered by
 scikit-learn (and the whole pydata stack, to be fair), the team in charge
 of the event had printed a huge scikit-learn logo behind me:
 https://twitter.com/GaelVaroquaux/status/1334959438059462659 (terrible
 mobile-phone picture)
 
 I would have liked to get a picture with the president and the logo, but
 it seems that they are releasing only a handful of pictures :(. Anyhow... 
 
 
 Thanks to the community! This is a huge success. For health topics (we
 are talking nationwide electronic health records) the ability to build on
 an independent open-source stack is extremely important. We, as a wider
 community, are building something priceless.
 
 Cheers,
 
 Ga?l
 _______________________________________________
 scikit-learn mailing list
 mailto:scikit-learn at python.org
 https://mail.python.org/mailman/listinfo/scikit-learn


_______________________________________________

scikit-learn mailing list 

mailto:scikit-learn at python.org 

https://mail.python.org/mailman/listinfo/scikit-learn
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mail.python.org/pipermail/scikit-learn/attachments/20201205/815765d4/attachment.html>

From solegalli at protonmail.com  Sat Dec  5 07:55:41 2020
From: solegalli at protonmail.com (Sole Galli)
Date: Sat, 05 Dec 2020 12:55:41 +0000
Subject: [scikit-learn] sample_weight vs class_weight
In-Reply-To: <CANThK3+3u3r5kJBCmJLs=k=_o00+YdKszqGY-yFCsd=-yRJ_oA@mail.gmail.com>
References: <gTEPm7FOtadRFt4Uh95vOMgl-21MkzmVmHKmpVMYV0g6GGuEWs7PoK355BXAFEFg8Fq6gRNbxJFJQxe3kyCqQDl60JdYuQ3pOxboYSHGwfI=@protonmail.com>
 <xHCo3nlA5pIjSkEFpDmhZXy0YnWvLhNH9Ev2V-yuhfBVAFyr-KEwcGeztAaR3MRWq1jM2VNOcNRRHBB5afeFzPwVz522sp7SNUHsTRSYhzc=@protonmail.com>
 <bfb529be-6de4-f6dc-ec68-821bade7ce25@gmail.com>
 <CANThK3+3u3r5kJBCmJLs=k=_o00+YdKszqGY-yFCsd=-yRJ_oA@mail.gmail.com>
Message-ID: <JLTumjG9PkLq7rW6gZ6dvv0LAbp225VRm-BY8HBUTTfhLdxqWtBWjA8aSn7HlunioKedM7kHR2q8QTl3febmQgojmU00WMIvuyfzWy6HuSM=@protonmail.com>

Thank you guys! very helpful :)

Soledad Galli
https://www.trainindata.com/

??????? Original Message ???????
On Friday, December 4, 2020 12:06 PM, mrschots <maykonschots at gmail.com> wrote:

> I have been using both in time-series classification. I put a exponential decay in sample_weights AND class weights as a dictionary.
>
> BR/Schots
>
> Em sex., 4 de dez. de 2020 ?s 12:01, Nicolas Hug <niourf at gmail.com> escreveu:
>
>> Basically passing class weights should be equivalent to passing per-class-constant sample weights.
>>
>>> why do some estimators allow to pass weights both as a dict in the init or as sample weights in fit? what's the logic?
>>
>> SW is a per-sample property (aligned with X and y) so we avoid passing those to init because the data isn't known when initializing the estimator. It's only known when calling fit. In general we avoid passing data-related info into init so that the same instance can be fitted on any data (with different number of samples, different classes, etc.).
>>
>> We allow to pass class_weight in init because the 'balanced' option is data-agnostic. Arguably, allowing a dict with actual class values violates the above argument (of not having data-related stuff in init), so I guess that's where the logic ends ;)
>>
>> As to why one would use both, I'm not so sure honestly.
>>
>> Nicolas
>>
>> On 12/4/20 10:40 AM, Sole Galli via scikit-learn wrote:
>>
>>> Actually, I found the answer. Both seem to be optimising the loss function for the various algorithms, below I include some links.
>>>
>>> If, we pass class_weight and sample_weight, then the final cost / weight is a combination of both.
>>>
>>> I have a follow up question: in which scenario would we use both? why do some estimators allow to pass weights both as a dict in the init or as sample weights in fit? what's the logic? I found it a bit confusing at the beginning.
>>>
>>> Thank you!
>>>
>>> https://stackoverflow.com/questions/30805192/scikit-learn-random-forest-class-weight-and-sample-weight-parameters
>>>
>>> https://stackoverflow.com/questions/30972029/how-does-the-class-weight-parameter-in-scikit-learn-work/30982811#30982811
>>>
>>> Soledad Galli
>>> https://www.trainindata.com/
>>>
>>> ??????? Original Message ???????
>>> On Thursday, December 3, 2020 11:55 AM, Sole Galli via scikit-learn [<scikit-learn at python.org>](mailto:scikit-learn at python.org) wrote:
>>>
>>>> Hello team,
>>>>
>>>> What is the difference in the implementation of class_weight and sample_weight in those algorithms that support both? like random forest or logistic regression?
>>>>
>>>> Are both modifying the loss function? in a similar way?
>>>>
>>>> Thank you!
>>>>
>>>> Sole
>>>
>>> _______________________________________________
>>> scikit-learn mailing list
>>> scikit-learn at python.org
>>>
>>> https://mail.python.org/mailman/listinfo/scikit-learn
>>
>> _______________________________________________
>> scikit-learn mailing list
>> scikit-learn at python.org
>> https://mail.python.org/mailman/listinfo/scikit-learn
>
> --
> Schots
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mail.python.org/pipermail/scikit-learn/attachments/20201205/db0a7f04/attachment-0001.html>

From jk231092 at gmail.com  Sat Dec  5 10:28:39 2020
From: jk231092 at gmail.com (Jitesh Khandelwal)
Date: Sat, 5 Dec 2020 20:58:39 +0530
Subject: [scikit-learn] Presented scikit-learn to the French President
In-Reply-To: <20201204210636.36geowufvvphujmg@phare.normalesup.org>
References: <20201204210636.36geowufvvphujmg@phare.normalesup.org>
Message-ID: <CAB+rR360+bzc991miq0q87HBn0EJ_GJK8RDPPHoYQ1wyv5ywSQ@mail.gmail.com>

Amazing, inspiring! Kudos to the sklearn team.

On Sat, Dec 5, 2020, 4:30 AM Gael Varoquaux <gael.varoquaux at normalesup.org>
wrote:

> Hi scikit-learn community,
>
> Today, I presented some efforts in digital health to the French president
> and part of the government. As these efforts were partly powered by
> scikit-learn (and the whole pydata stack, to be fair), the team in charge
> of the event had printed a huge scikit-learn logo behind me:
> https://twitter.com/GaelVaroquaux/status/1334959438059462659 (terrible
> mobile-phone picture)
>
> I would have liked to get a picture with the president and the logo, but
> it seems that they are releasing only a handful of pictures :(. Anyhow...
>
>
> Thanks to the community! This is a huge success. For health topics (we
> are talking nationwide electronic health records) the ability to build on
> an independent open-source stack is extremely important. We, as a wider
> community, are building something priceless.
>
> Cheers,
>
> Ga?l
> _______________________________________________
> scikit-learn mailing list
> scikit-learn at python.org
> https://mail.python.org/mailman/listinfo/scikit-learn
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mail.python.org/pipermail/scikit-learn/attachments/20201205/33f77751/attachment.html>

From mail at sebastianraschka.com  Sat Dec  5 11:05:48 2020
From: mail at sebastianraschka.com (Sebastian Raschka)
Date: Sat, 5 Dec 2020 10:05:48 -0600
Subject: [scikit-learn] Presented scikit-learn to the French President
In-Reply-To: <CAB+rR360+bzc991miq0q87HBn0EJ_GJK8RDPPHoYQ1wyv5ywSQ@mail.gmail.com>
References: <20201204210636.36geowufvvphujmg@phare.normalesup.org>
 <CAB+rR360+bzc991miq0q87HBn0EJ_GJK8RDPPHoYQ1wyv5ywSQ@mail.gmail.com>
Message-ID: <48B89C62-8C0E-4B8C-BB4E-58BAD976F684@sebastianraschka.com>

This is really awesome news! Thanks a lot to everyone developing scikit-learn. I am just wrapping up another successful semester, teaching students ML basics. Most coming from an R background, they really loved scikit-learn and appreciated it's ease of use and well-thought-out API.

Best,
Sebastian

> On Dec 5, 2020, at 9:28 AM, Jitesh Khandelwal <jk231092 at gmail.com> wrote:
> 
> Amazing, inspiring! Kudos to the sklearn team.
> 
> On Sat, Dec 5, 2020, 4:30 AM Gael Varoquaux <gael.varoquaux at normalesup.org> wrote:
> Hi scikit-learn community,
> 
> Today, I presented some efforts in digital health to the French president
> and part of the government. As these efforts were partly powered by
> scikit-learn (and the whole pydata stack, to be fair), the team in charge
> of the event had printed a huge scikit-learn logo behind me:
> https://twitter.com/GaelVaroquaux/status/1334959438059462659 (terrible
> mobile-phone picture)
> 
> I would have liked to get a picture with the president and the logo, but
> it seems that they are releasing only a handful of pictures :(. Anyhow... 
> 
> 
> Thanks to the community! This is a huge success. For health topics (we
> are talking nationwide electronic health records) the ability to build on
> an independent open-source stack is extremely important. We, as a wider
> community, are building something priceless.
> 
> Cheers,
> 
> Ga?l
> _______________________________________________
> scikit-learn mailing list
> scikit-learn at python.org
> https://mail.python.org/mailman/listinfo/scikit-learn
> _______________________________________________
> scikit-learn mailing list
> scikit-learn at python.org
> https://mail.python.org/mailman/listinfo/scikit-learn


From jbbrown at kuhp.kyoto-u.ac.jp  Sun Dec  6 16:36:42 2020
From: jbbrown at kuhp.kyoto-u.ac.jp (Brown J.B.)
Date: Sun, 6 Dec 2020 22:36:42 +0100
Subject: [scikit-learn] Presented scikit-learn to the French President
In-Reply-To: <48B89C62-8C0E-4B8C-BB4E-58BAD976F684@sebastianraschka.com>
References: <20201204210636.36geowufvvphujmg@phare.normalesup.org>
 <CAB+rR360+bzc991miq0q87HBn0EJ_GJK8RDPPHoYQ1wyv5ywSQ@mail.gmail.com>
 <48B89C62-8C0E-4B8C-BB4E-58BAD976F684@sebastianraschka.com>
Message-ID: <CAJe_vxDKycXVqj=O4PYOwJB+PH5dTJZaYWkb65qVmiu0pBwXHg@mail.gmail.com>

Congratulations to all developers and contributors to scikit-learn, from
core-devs to webmasters, documentation checkers and commenters, and other
facilitators!
Keeping a project alive takes a substantial amount of vision and hard work,
and scikit-learn is a mature ecosystem because of the vision and hard work
of everyone.

This recognition by the French government is fantastic -- congratulations
Gael to you, your leadership, and your team!
In fact, scikit-learn is probably more ubiquitous than anyone individually
recognizes, because for all of the contributions in github and mailing
lists, there are probably many more people who are benefitting from
applying it to their individual scenarios.

I myself am a very appreciative user. :)
Sincere regards and congratulations again,
J.B. Brown



2020?12?5?(?) 17:53 Sebastian Raschka <mail at sebastianraschka.com>:

> This is really awesome news! Thanks a lot to everyone developing
> scikit-learn. I am just wrapping up another successful semester, teaching
> students ML basics. Most coming from an R background, they really loved
> scikit-learn and appreciated it's ease of use and well-thought-out API.
>
> Best,
> Sebastian
>
> > On Dec 5, 2020, at 9:28 AM, Jitesh Khandelwal <jk231092 at gmail.com>
> wrote:
> >
> > Amazing, inspiring! Kudos to the sklearn team.
> >
> > On Sat, Dec 5, 2020, 4:30 AM Gael Varoquaux <
> gael.varoquaux at normalesup.org> wrote:
> > Hi scikit-learn community,
> >
> > Today, I presented some efforts in digital health to the French president
> > and part of the government. As these efforts were partly powered by
> > scikit-learn (and the whole pydata stack, to be fair), the team in charge
> > of the event had printed a huge scikit-learn logo behind me:
> > https://twitter.com/GaelVaroquaux/status/1334959438059462659 (terrible
> > mobile-phone picture)
> >
> > I would have liked to get a picture with the president and the logo, but
> > it seems that they are releasing only a handful of pictures :(.
> Anyhow...
> >
> >
> > Thanks to the community! This is a huge success. For health topics (we
> > are talking nationwide electronic health records) the ability to build on
> > an independent open-source stack is extremely important. We, as a wider
> > community, are building something priceless.
> >
> > Cheers,
> >
> > Ga?l
> > _______________________________________________
> > scikit-learn mailing list
> > scikit-learn at python.org
> > https://mail.python.org/mailman/listinfo/scikit-learn
> > _______________________________________________
> > scikit-learn mailing list
> > scikit-learn at python.org
> > https://mail.python.org/mailman/listinfo/scikit-learn
>
> _______________________________________________
> scikit-learn mailing list
> scikit-learn at python.org
> https://mail.python.org/mailman/listinfo/scikit-learn
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mail.python.org/pipermail/scikit-learn/attachments/20201206/40c788f0/attachment.html>

From adrin.jalali at gmail.com  Tue Dec  8 07:01:56 2020
From: adrin.jalali at gmail.com (Adrin)
Date: Tue, 8 Dec 2020 13:01:56 +0100
Subject: [scikit-learn] Changes in Travis billing
In-Reply-To: <CAEjJ3fhMtFe3JaQnusA12xFTE=Y3mGdBr+U=3vs3nxwnoaHWtA@mail.gmail.com>
References: <20201102105018.kfarylwpao6iliju@phare.normalesup.org>
 <CAEOrW49QWQLKi0Rd5AdESD=YCwsm356egRdH+=fxtqZeiMZNzw@mail.gmail.com>
 <CAFvE7K6dPVtH4e6=0toWujbLiByg=mVOxayh3rTvEUNZCt8q+g@mail.gmail.com>
 <CAEOrW49SNzEffyWNOLTahokJiGQoFCM7bx+s5GYjMrNnw2ZrFg@mail.gmail.com>
 <20201126140652.tdrsjzfasb64kqua@phare.normalesup.org>
 <20201126141212.x72eecggoozdocck@phare.normalesup.org>
 <emdf0ee827-9920-429c-ad0a-d15556e8557a@desktop-fo3usmm>
 <CACDxx9gSGfoKOQMds2pkmSKVFR2eAkP3EVmxrLWPrwu4u5hwMg@mail.gmail.com>
 <CAEOrW4-437koMoQ=Y-uT4W7Cp74Wg4n9Qktm=CM3f02142vXuQ@mail.gmail.com>
 <CAE-UAvTdYFZE+rvQTVFtCAmoafJea-dHsf12m+jf+1duAVkvxg@mail.gmail.com>
 <CAEjJ3fhMtFe3JaQnusA12xFTE=Y3mGdBr+U=3vs3nxwnoaHWtA@mail.gmail.com>
Message-ID: <CAEOrW4-tu0EG6C8+w6TifBke-5oBMC0v2128ZHe5bxNo5MrKdg@mail.gmail.com>

We've got 10k extra credit, not sure if this is monthly or a one time one.

On Thu, Dec 3, 2020 at 9:00 PM Brigitta Sipocz <bsipocz at gmail.com> wrote:

> Hi,
>
> Astropy also got a (one time) credit, that we very quickly burned up. And
> NumPy also got a credit, but I'm not sure whether it's also a one-off or a
> monthly.
>
> For astropy, our workaround is to run the tests on aarch64 and the other
> more exotic hardware from a weekly cron, as each takes a very long time on
> GH Actions. Hopefully, a better, native solution will come to Actions/Azure
> soon enough.
> I link the PR that set this up, maybe this solution would be good enough
> for sklearn, too.
>
> https://github.com/astropy/astropy/pull/11045
>
>
> Cheers,
>  Brigitta
>
> On Thu, 3 Dec 2020 at 02:50, Nelle Varoquaux <nelle.varoquaux at gmail.com>
> wrote:
>
>> Wow? what?! That's insane?
>> scikit-image got some credits (
>> https://github.com/scikit-image/scikit-image-wheels/pull/47#issuecomment-736760539)
>> without much issue? Maybe someone should reach out directly on this thread
>> to the travis people?
>>
>> Cheers,
>> N
>>
>> On Thu, 3 Dec 2020 at 11:37, Adrin <adrin.jalali at gmail.com> wrote:
>>
>>> I got a response from Travis, and according to this, we don't qualify
>>> for the free credits:
>>>
>>> The free plan will grant your organization 10000 credits.
>>>
>>> We offer an Open Source Subscription for free to non-commercial
>>> open-source projects. To qualify for an Open Source subscription, the
>>> project must meet the following requirements:
>>>
>>>
>>>    - You are a project lead or regular committer (latest commit in the
>>>    last month)
>>>    - Project must be at least 3 months old and is in active development
>>>    (with regular commits and activity)
>>>    - Project meets the OSD <https://opensource.org/docs/osd>
>>>    specification
>>>    -
>>> *Project must not be sponsored by a commercial company or organization
>>>    (monetary or with employees paid to work on the project) *
>>>    - Project can not provide commercial services or distribute paid
>>>    versions of the software
>>>
>>>
>>> Does this sound like you and your project? We'd be very happy to support
>>> you!
>>>
>>> However, if your project does not match these requirements or you have
>>> further questions [1], please feel free to ask!
>>>
>>> We look forward to your response if you meet these requirements to
>>> proceed with the next steps.
>>>
>>> Thank you
>>>
>>> On Thu, Dec 3, 2020 at 10:17 AM Guillaume Lema?tre <
>>> g.lemaitre58 at gmail.com> wrote:
>>>
>>>> ARM support
>>>>
>>>> On Thu, 3 Dec 2020 at 01:37, Andreas C. Mueller <
>>>> andreasmuellerml at gmail.com> wrote:
>>>>
>>>>> Sorry I'm probably missing some detail but what does travis provide
>>>>> that
>>>>> github actions and azure pipeline don't provide?
>>>>>
>>>>> ------ Original Message ------
>>>>> From: "Gael Varoquaux" <gael.varoquaux at normalesup.org>
>>>>> To: "Scikit-learn mailing list" <scikit-learn at python.org>
>>>>> Sent: 11/26/2020 6:12:12 AM
>>>>> Subject: Re: [scikit-learn] Changes in Travis billing
>>>>>
>>>>> >On Thu, Nov 26, 2020 at 03:06:52PM +0100, Gael Varoquaux wrote:
>>>>> >>  On Thu, Nov 26, 2020 at 02:45:33PM +0100, Adrin wrote:
>>>>> >>  > At this point I'm at a loss, and reading the NumFocus chat and
>>>>> other
>>>>> >>  > packages' experience with them on the same topic, seems like we
>>>>> just
>>>>> >>  > need to move out of Travis.
>>>>> >
>>>>> >>  Agreed. Do we still need them for something essential?
>>>>> >
>>>>> >Sorry, ARM, it was just above in the conversation.
>>>>> >
>>>>> >I think that we have no other option than reduce the frequency of the
>>>>> >cron, and wait for other platforms to offer ARM, which will hopefully
>>>>> >happen soonish.
>>>>> >
>>>>> >G
>>>>> >_______________________________________________
>>>>> >scikit-learn mailing list
>>>>> >scikit-learn at python.org
>>>>> >https://mail.python.org/mailman/listinfo/scikit-learn
>>>>>
>>>>> _______________________________________________
>>>>> scikit-learn mailing list
>>>>> scikit-learn at python.org
>>>>> https://mail.python.org/mailman/listinfo/scikit-learn
>>>>>
>>>>
>>>>
>>>> --
>>>> Guillaume Lemaitre
>>>> Scikit-learn @ Inria Foundation
>>>> https://glemaitre.github.io/
>>>> _______________________________________________
>>>> scikit-learn mailing list
>>>> scikit-learn at python.org
>>>> https://mail.python.org/mailman/listinfo/scikit-learn
>>>>
>>> _______________________________________________
>>> scikit-learn mailing list
>>> scikit-learn at python.org
>>> https://mail.python.org/mailman/listinfo/scikit-learn
>>>
>> _______________________________________________
>> scikit-learn mailing list
>> scikit-learn at python.org
>> https://mail.python.org/mailman/listinfo/scikit-learn
>>
> _______________________________________________
> scikit-learn mailing list
> scikit-learn at python.org
> https://mail.python.org/mailman/listinfo/scikit-learn
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mail.python.org/pipermail/scikit-learn/attachments/20201208/4bf01c4b/attachment.html>

From mahmood.nt at gmail.com  Wed Dec  9 04:39:14 2020
From: mahmood.nt at gmail.com (Mahmood Naderan)
Date: Wed, 9 Dec 2020 10:39:14 +0100
Subject: [scikit-learn] Drawing contours in KMeans
Message-ID: <CADa2P2VP0sDWyizfEoGm+qfT9dssQoDn2CMGA+xfp+16nnmbUQ@mail.gmail.com>

Hi
I use the following code to highlight the cluster centers with some red
dots.

kmeans = KMeans(n_clusters=6, init='k-means++', max_iter=100, n_init=10,
random_state=0)
pred_y = kmeans.fit_predict(a)
plt.scatter(a[:,0], a[:,1])
plt.scatter(kmeans.cluster_centers_[:, 0], kmeans.cluster_centers_[:, 1],
s=100, c='red')
plt.show()

I would like to know if it is possible to draw contours over the clusters.
Is there any way for that?
Please let me know if there is a function or option in KMeans.

Regards,
Mahmood
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mail.python.org/pipermail/scikit-learn/attachments/20201209/78482f46/attachment.html>

From ahowe42 at gmail.com  Wed Dec  9 05:47:40 2020
From: ahowe42 at gmail.com (Andrew Howe)
Date: Wed, 9 Dec 2020 10:47:40 +0000
Subject: [scikit-learn] Drawing contours in KMeans
In-Reply-To: <CADa2P2VP0sDWyizfEoGm+qfT9dssQoDn2CMGA+xfp+16nnmbUQ@mail.gmail.com>
References: <CADa2P2VP0sDWyizfEoGm+qfT9dssQoDn2CMGA+xfp+16nnmbUQ@mail.gmail.com>
Message-ID: <CANnYi3Q+nsF2NzeWPxTSqn+ugRbPJktRMjeLhuGds=Tr0c+kRQ@mail.gmail.com>

Contours generally indicate a third variable - often a probability density.
Kmeans doesn't provide density estimates, so what precisely would you want
the contours to represent?

Andrew

<~~~~~~~~~~~~~~~~~~~~~~~~~~~>
J. Andrew Howe, PhD
LinkedIn Profile <http://www.linkedin.com/in/ahowe42>
ResearchGate Profile <http://www.researchgate.net/profile/John_Howe12/>
Open Researcher and Contributor ID (ORCID)
<http://orcid.org/0000-0002-3553-1990>
Github Profile <http://github.com/ahowe42>
Personal Website <http://www.andrewhowe.com>
I live to learn, so I can learn to live. - me
<~~~~~~~~~~~~~~~~~~~~~~~~~~~>


On Wed, Dec 9, 2020 at 9:41 AM Mahmood Naderan <mahmood.nt at gmail.com> wrote:

> Hi
> I use the following code to highlight the cluster centers with some red
> dots.
>
> kmeans = KMeans(n_clusters=6, init='k-means++', max_iter=100, n_init=10,
> random_state=0)
> pred_y = kmeans.fit_predict(a)
> plt.scatter(a[:,0], a[:,1])
> plt.scatter(kmeans.cluster_centers_[:, 0], kmeans.cluster_centers_[:, 1],
> s=100, c='red')
> plt.show()
>
> I would like to know if it is possible to draw contours over the clusters.
> Is there any way for that?
> Please let me know if there is a function or option in KMeans.
>
> Regards,
> Mahmood
>
>
> _______________________________________________
> scikit-learn mailing list
> scikit-learn at python.org
> https://mail.python.org/mailman/listinfo/scikit-learn
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mail.python.org/pipermail/scikit-learn/attachments/20201209/77f37591/attachment.html>

From mahmood.nt at gmail.com  Wed Dec  9 07:57:18 2020
From: mahmood.nt at gmail.com (Mahmood Naderan)
Date: Wed, 9 Dec 2020 13:57:18 +0100
Subject: [scikit-learn] Drawing contours in KMeans
In-Reply-To: <CANnYi3Q+nsF2NzeWPxTSqn+ugRbPJktRMjeLhuGds=Tr0c+kRQ@mail.gmail.com>
References: <CADa2P2VP0sDWyizfEoGm+qfT9dssQoDn2CMGA+xfp+16nnmbUQ@mail.gmail.com>
 <CANnYi3Q+nsF2NzeWPxTSqn+ugRbPJktRMjeLhuGds=Tr0c+kRQ@mail.gmail.com>
Message-ID: <CADa2P2X1jvzwAWwtGJoZoe2=gq1JPgCj6n7FNNtG-EZB+qmpcg@mail.gmail.com>

I mean a circle/contour to group the points in a cluster for better
representation.
For example, if there are 6 six clusters, it will be more meaningful to
group large data points in a circle or contour.

Regards,
Mahmood




On Wed, Dec 9, 2020 at 11:49 AM Andrew Howe <ahowe42 at gmail.com> wrote:

> Contours generally indicate a third variable - often a probability
> density. Kmeans doesn't provide density estimates, so what precisely would
> you want the contours to represent?
>
> Andrew
>
> <~~~~~~~~~~~~~~~~~~~~~~~~~~~>
> J. Andrew Howe, PhD
> LinkedIn Profile <http://www.linkedin.com/in/ahowe42>
> ResearchGate Profile <http://www.researchgate.net/profile/John_Howe12/>
> Open Researcher and Contributor ID (ORCID)
> <http://orcid.org/0000-0002-3553-1990>
> Github Profile <http://github.com/ahowe42>
> Personal Website <http://www.andrewhowe.com>
> I live to learn, so I can learn to live. - me
> <~~~~~~~~~~~~~~~~~~~~~~~~~~~>
>
>
> On Wed, Dec 9, 2020 at 9:41 AM Mahmood Naderan <mahmood.nt at gmail.com>
> wrote:
>
>> Hi
>> I use the following code to highlight the cluster centers with some red
>> dots.
>>
>> kmeans = KMeans(n_clusters=6, init='k-means++', max_iter=100, n_init=10,
>> random_state=0)
>> pred_y = kmeans.fit_predict(a)
>> plt.scatter(a[:,0], a[:,1])
>> plt.scatter(kmeans.cluster_centers_[:, 0], kmeans.cluster_centers_[:, 1],
>> s=100, c='red')
>> plt.show()
>>
>> I would like to know if it is possible to draw contours over the
>> clusters. Is there any way for that?
>> Please let me know if there is a function or option in KMeans.
>>
>> Regards,
>> Mahmood
>>
>>
>> _______________________________________________
>> scikit-learn mailing list
>> scikit-learn at python.org
>> https://mail.python.org/mailman/listinfo/scikit-learn
>>
> _______________________________________________
> scikit-learn mailing list
> scikit-learn at python.org
> https://mail.python.org/mailman/listinfo/scikit-learn
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mail.python.org/pipermail/scikit-learn/attachments/20201209/2aa3b0f0/attachment.html>

From ahowe42 at gmail.com  Wed Dec  9 14:22:30 2020
From: ahowe42 at gmail.com (Andrew Howe)
Date: Wed, 9 Dec 2020 19:22:30 +0000
Subject: [scikit-learn] Drawing contours in KMeans
In-Reply-To: <CADa2P2X1jvzwAWwtGJoZoe2=gq1JPgCj6n7FNNtG-EZB+qmpcg@mail.gmail.com>
References: <CADa2P2VP0sDWyizfEoGm+qfT9dssQoDn2CMGA+xfp+16nnmbUQ@mail.gmail.com>
 <CANnYi3Q+nsF2NzeWPxTSqn+ugRbPJktRMjeLhuGds=Tr0c+kRQ@mail.gmail.com>
 <CADa2P2X1jvzwAWwtGJoZoe2=gq1JPgCj6n7FNNtG-EZB+qmpcg@mail.gmail.com>
Message-ID: <CANnYi3RvfB6mYxxmXw2XqagjW6dXLQdWnebWYSsVm3OcBkwQgg@mail.gmail.com>

Ok, I see. Well the attached notebook demonstrates doing this by simply
finding the maximum distance from each centroid to it's datapoints and
drawing a circle using that radius. It's simple, but will hopefully at
least point you in a useful direction.
[image: image.png]
Andrew

<~~~~~~~~~~~~~~~~~~~~~~~~~~~>
J. Andrew Howe, PhD
LinkedIn Profile <http://www.linkedin.com/in/ahowe42>
ResearchGate Profile <http://www.researchgate.net/profile/John_Howe12/>
Open Researcher and Contributor ID (ORCID)
<http://orcid.org/0000-0002-3553-1990>
Github Profile <http://github.com/ahowe42>
Personal Website <http://www.andrewhowe.com>
I live to learn, so I can learn to live. - me
<~~~~~~~~~~~~~~~~~~~~~~~~~~~>


On Wed, Dec 9, 2020 at 12:59 PM Mahmood Naderan <mahmood.nt at gmail.com>
wrote:

> I mean a circle/contour to group the points in a cluster for better
> representation.
> For example, if there are 6 six clusters, it will be more meaningful to
> group large data points in a circle or contour.
>
> Regards,
> Mahmood
>
>
>
>
> On Wed, Dec 9, 2020 at 11:49 AM Andrew Howe <ahowe42 at gmail.com> wrote:
>
>> Contours generally indicate a third variable - often a probability
>> density. Kmeans doesn't provide density estimates, so what precisely would
>> you want the contours to represent?
>>
>> Andrew
>>
>> <~~~~~~~~~~~~~~~~~~~~~~~~~~~>
>> J. Andrew Howe, PhD
>> LinkedIn Profile <http://www.linkedin.com/in/ahowe42>
>> ResearchGate Profile <http://www.researchgate.net/profile/John_Howe12/>
>> Open Researcher and Contributor ID (ORCID)
>> <http://orcid.org/0000-0002-3553-1990>
>> Github Profile <http://github.com/ahowe42>
>> Personal Website <http://www.andrewhowe.com>
>> I live to learn, so I can learn to live. - me
>> <~~~~~~~~~~~~~~~~~~~~~~~~~~~>
>>
>>
>> On Wed, Dec 9, 2020 at 9:41 AM Mahmood Naderan <mahmood.nt at gmail.com>
>> wrote:
>>
>>> Hi
>>> I use the following code to highlight the cluster centers with some red
>>> dots.
>>>
>>> kmeans = KMeans(n_clusters=6, init='k-means++', max_iter=100, n_init=10,
>>> random_state=0)
>>> pred_y = kmeans.fit_predict(a)
>>> plt.scatter(a[:,0], a[:,1])
>>> plt.scatter(kmeans.cluster_centers_[:, 0], kmeans.cluster_centers_[:,
>>> 1], s=100, c='red')
>>> plt.show()
>>>
>>> I would like to know if it is possible to draw contours over the
>>> clusters. Is there any way for that?
>>> Please let me know if there is a function or option in KMeans.
>>>
>>> Regards,
>>> Mahmood
>>>
>>>
>>> _______________________________________________
>>> scikit-learn mailing list
>>> scikit-learn at python.org
>>> https://mail.python.org/mailman/listinfo/scikit-learn
>>>
>> _______________________________________________
>> scikit-learn mailing list
>> scikit-learn at python.org
>> https://mail.python.org/mailman/listinfo/scikit-learn
>>
> _______________________________________________
> scikit-learn mailing list
> scikit-learn at python.org
> https://mail.python.org/mailman/listinfo/scikit-learn
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mail.python.org/pipermail/scikit-learn/attachments/20201209/8ea56d53/attachment-0001.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: image.png
Type: image/png
Size: 44525 bytes
Desc: not available
URL: <https://mail.python.org/pipermail/scikit-learn/attachments/20201209/8ea56d53/attachment-0001.png>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: KMeans_Cluster_Circle.ipynb
Type: application/octet-stream
Size: 27501 bytes
Desc: not available
URL: <https://mail.python.org/pipermail/scikit-learn/attachments/20201209/8ea56d53/attachment-0001.obj>

From helmrp at yahoo.com  Wed Dec  9 14:53:06 2020
From: helmrp at yahoo.com (The Helmbolds)
Date: Wed, 9 Dec 2020 19:53:06 +0000 (UTC)
Subject: [scikit-learn] Drawing contours in KMeans
In-Reply-To: <CANnYi3RvfB6mYxxmXw2XqagjW6dXLQdWnebWYSsVm3OcBkwQgg@mail.gmail.com>
References: <CADa2P2VP0sDWyizfEoGm+qfT9dssQoDn2CMGA+xfp+16nnmbUQ@mail.gmail.com>
 <CANnYi3Q+nsF2NzeWPxTSqn+ugRbPJktRMjeLhuGds=Tr0c+kRQ@mail.gmail.com>
 <CADa2P2X1jvzwAWwtGJoZoe2=gq1JPgCj6n7FNNtG-EZB+qmpcg@mail.gmail.com>
 <CANnYi3RvfB6mYxxmXw2XqagjW6dXLQdWnebWYSsVm3OcBkwQgg@mail.gmail.com>
Message-ID: <694862547.3937714.1607543587026@mail.yahoo.com>

[scikit-learn] Drawing contours in KMeans4

Mebbe principal components analysis would suggest an ellipsoid containing "most" of the points in a "cloud".



"You won't find the right answers if you don't ask the right questions!" (Robert Helmbold, 2013) 

    On Wednesday, December 9, 2020, 12:22:49 PM MST, Andrew Howe <ahowe42 at gmail.com> wrote:  
 
 Ok, I see. Well the attached notebook demonstrates?doing this by simply finding the maximum distance from each centroid to it's datapoints?and drawing a circle using that radius. It's simple, but will hopefully at least point you?in a useful direction.
Andrew
<~~~~~~~~~~~~~~~~~~~~~~~~~~~>
J. Andrew Howe, PhDLinkedIn ProfileResearchGate ProfileOpen Researcher and Contributor ID (ORCID)Github Profile
Personal WebsiteI live to learn, so I can learn to live. - me
<~~~~~~~~~~~~~~~~~~~~~~~~~~~>

On Wed, Dec 9, 2020 at 12:59 PM Mahmood Naderan <mahmood.nt at gmail.com> wrote:

I mean a circle/contour to group the points in a cluster for better representation. 
For example, if there are 6 six clusters, it will be more meaningful to group large data points in a circle or contour.
Regards,
Mahmood




On Wed, Dec 9, 2020 at 11:49 AM Andrew Howe <ahowe42 at gmail.com> wrote:

Contours generally indicate a third variable - often a probability density. Kmeans doesn't provide density estimates, so what precisely would you want the contours to represent?
Andrew
<~~~~~~~~~~~~~~~~~~~~~~~~~~~>
J. Andrew Howe, PhDLinkedIn ProfileResearchGate ProfileOpen Researcher and Contributor ID (ORCID)Github Profile
Personal WebsiteI live to learn, so I can learn to live. - me
<~~~~~~~~~~~~~~~~~~~~~~~~~~~>

On Wed, Dec 9, 2020 at 9:41 AM Mahmood Naderan <mahmood.nt at gmail.com> wrote:

HiI use the following code to highlight the cluster centers with some red dots.
kmeans = KMeans(n_clusters=6, init='k-means++', max_iter=100, n_init=10, random_state=0)
pred_y = kmeans.fit_predict(a)
plt.scatter(a[:,0], a[:,1])
plt.scatter(kmeans.cluster_centers_[:, 0], kmeans.cluster_centers_[:, 1], s=100, c='red')
plt.show()

I would like to know if it is possible to draw contours over the clusters. Is there any way for that?Please let me know if there is a function or option in KMeans.
Regards,
Mahmood


_______________________________________________
scikit-learn mailing list
scikit-learn at python.org
https://mail.python.org/mailman/listinfo/scikit-learn

_______________________________________________
scikit-learn mailing list
scikit-learn at python.org
https://mail.python.org/mailman/listinfo/scikit-learn

_______________________________________________
scikit-learn mailing list
scikit-learn at python.org
https://mail.python.org/mailman/listinfo/scikit-learn

_______________________________________________
scikit-learn mailing list
scikit-learn at python.org
https://mail.python.org/mailman/listinfo/scikit-learn
  
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mail.python.org/pipermail/scikit-learn/attachments/20201209/d56d1184/attachment-0001.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: image.png
Type: image/png
Size: 44525 bytes
Desc: not available
URL: <https://mail.python.org/pipermail/scikit-learn/attachments/20201209/d56d1184/attachment-0001.png>

From mahmood.nt at gmail.com  Wed Dec  9 15:06:19 2020
From: mahmood.nt at gmail.com (Mahmood Naderan)
Date: Wed, 9 Dec 2020 21:06:19 +0100
Subject: [scikit-learn] Drawing contours in KMeans
In-Reply-To: <694862547.3937714.1607543587026@mail.yahoo.com>
References: <CADa2P2VP0sDWyizfEoGm+qfT9dssQoDn2CMGA+xfp+16nnmbUQ@mail.gmail.com>
 <CANnYi3Q+nsF2NzeWPxTSqn+ugRbPJktRMjeLhuGds=Tr0c+kRQ@mail.gmail.com>
 <CADa2P2X1jvzwAWwtGJoZoe2=gq1JPgCj6n7FNNtG-EZB+qmpcg@mail.gmail.com>
 <CANnYi3RvfB6mYxxmXw2XqagjW6dXLQdWnebWYSsVm3OcBkwQgg@mail.gmail.com>
 <694862547.3937714.1607543587026@mail.yahoo.com>
Message-ID: <CADa2P2WbF=dVz3dQjtMzqEfutC9KrGOm9z_iOJrJXZQ1Sf7sMg@mail.gmail.com>

 >Mebbe principal components analysis would suggest an
>ellipsoid containing "most" of the points in a "cloud".

Sorry I didn't understand. Can you explain more?
Regards,
Mahmood




On Wed, Dec 9, 2020 at 8:55 PM The Helmbolds via scikit-learn <
scikit-learn at python.org> wrote:

> [scikit-learn] Drawing contours in KMeans4
>
>
> Mebbe principal components analysis would suggest an ellipsoid containing
> "most" of the points in a "cloud".
>
>
>
>
> "You won't find the right answers if you don't ask the right questions!"
> (Robert Helmbold, 2013)
>
>
> On Wednesday, December 9, 2020, 12:22:49 PM MST, Andrew Howe <
> ahowe42 at gmail.com> wrote:
>
>
> Ok, I see. Well the attached notebook demonstrates doing this by simply
> finding the maximum distance from each centroid to it's datapoints and
> drawing a circle using that radius. It's simple, but will hopefully at
> least point you in a useful direction.
> [image: image.png]
> Andrew
>
> <~~~~~~~~~~~~~~~~~~~~~~~~~~~>
> J. Andrew Howe, PhD
> LinkedIn Profile <http://www.linkedin.com/in/ahowe42>
> ResearchGate Profile <http://www.researchgate.net/profile/John_Howe12/>
> Open Researcher and Contributor ID (ORCID)
> <http://orcid.org/0000-0002-3553-1990>
> Github Profile <http://github.com/ahowe42>
> Personal Website <http://www.andrewhowe.com>
> I live to learn, so I can learn to live. - me
> <~~~~~~~~~~~~~~~~~~~~~~~~~~~>
>
>
> On Wed, Dec 9, 2020 at 12:59 PM Mahmood Naderan <mahmood.nt at gmail.com>
> wrote:
>
> I mean a circle/contour to group the points in a cluster for better
> representation.
> For example, if there are 6 six clusters, it will be more meaningful to
> group large data points in a circle or contour.
>
> Regards,
> Mahmood
>
>
>
>
> On Wed, Dec 9, 2020 at 11:49 AM Andrew Howe <ahowe42 at gmail.com> wrote:
>
> Contours generally indicate a third variable - often a probability
> density. Kmeans doesn't provide density estimates, so what precisely would
> you want the contours to represent?
>
> Andrew
>
> <~~~~~~~~~~~~~~~~~~~~~~~~~~~>
> J. Andrew Howe, PhD
> LinkedIn Profile <http://www.linkedin.com/in/ahowe42>
> ResearchGate Profile <http://www.researchgate.net/profile/John_Howe12/>
> Open Researcher and Contributor ID (ORCID)
> <http://orcid.org/0000-0002-3553-1990>
> Github Profile <http://github.com/ahowe42>
> Personal Website <http://www.andrewhowe.com>
> I live to learn, so I can learn to live. - me
> <~~~~~~~~~~~~~~~~~~~~~~~~~~~>
>
>
> On Wed, Dec 9, 2020 at 9:41 AM Mahmood Naderan <mahmood.nt at gmail.com>
> wrote:
>
> Hi
> I use the following code to highlight the cluster centers with some red
> dots.
>
> kmeans = KMeans(n_clusters=6, init='k-means++', max_iter=100, n_init=10,
> random_state=0)
> pred_y = kmeans.fit_predict(a)
> plt.scatter(a[:,0], a[:,1])
> plt.scatter(kmeans.cluster_centers_[:, 0], kmeans.cluster_centers_[:, 1],
> s=100, c='red')
> plt.show()
>
> I would like to know if it is possible to draw contours over the clusters.
> Is there any way for that?
> Please let me know if there is a function or option in KMeans.
>
> Regards,
> Mahmood
>
>
> _______________________________________________
> scikit-learn mailing list
> scikit-learn at python.org
> https://mail.python.org/mailman/listinfo/scikit-learn
>
> _______________________________________________
> scikit-learn mailing list
> scikit-learn at python.org
> https://mail.python.org/mailman/listinfo/scikit-learn
>
> _______________________________________________
> scikit-learn mailing list
> scikit-learn at python.org
> https://mail.python.org/mailman/listinfo/scikit-learn
>
> _______________________________________________
> scikit-learn mailing list
> scikit-learn at python.org
> https://mail.python.org/mailman/listinfo/scikit-learn
> _______________________________________________
> scikit-learn mailing list
> scikit-learn at python.org
> https://mail.python.org/mailman/listinfo/scikit-learn
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mail.python.org/pipermail/scikit-learn/attachments/20201209/7f22bf3a/attachment-0001.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: image.png
Type: image/png
Size: 44525 bytes
Desc: not available
URL: <https://mail.python.org/pipermail/scikit-learn/attachments/20201209/7f22bf3a/attachment-0001.png>

From jbbrown at kuhp.kyoto-u.ac.jp  Wed Dec  9 15:40:15 2020
From: jbbrown at kuhp.kyoto-u.ac.jp (Brown J.B.)
Date: Wed, 9 Dec 2020 21:40:15 +0100
Subject: [scikit-learn] Drawing contours in KMeans
In-Reply-To: <CADa2P2WbF=dVz3dQjtMzqEfutC9KrGOm9z_iOJrJXZQ1Sf7sMg@mail.gmail.com>
References: <CADa2P2VP0sDWyizfEoGm+qfT9dssQoDn2CMGA+xfp+16nnmbUQ@mail.gmail.com>
 <CANnYi3Q+nsF2NzeWPxTSqn+ugRbPJktRMjeLhuGds=Tr0c+kRQ@mail.gmail.com>
 <CADa2P2X1jvzwAWwtGJoZoe2=gq1JPgCj6n7FNNtG-EZB+qmpcg@mail.gmail.com>
 <CANnYi3RvfB6mYxxmXw2XqagjW6dXLQdWnebWYSsVm3OcBkwQgg@mail.gmail.com>
 <694862547.3937714.1607543587026@mail.yahoo.com>
 <CADa2P2WbF=dVz3dQjtMzqEfutC9KrGOm9z_iOJrJXZQ1Sf7sMg@mail.gmail.com>
Message-ID: <CAJe_vxBnwB_iR4PbX-ADX4DS1oGc4E=VcJS=6mvVB8f5t6H-gA@mail.gmail.com>

Dear Mahmood,

Andrew's solution with a circle will guarantee you render an image in which
every point is covered within some circle.

However, if data contains outliers or artifacts, you might get circles
which are excessively large and distort the image you want.
For example, imagine if there were a single red point in Andrew's image at
the coordinate (3,10); then, the resulting circle would cover all points in
the entire plot, which is unlikely what you want.
You could potentially generate a density estimate for each class and then
have matplotlib render the contour lines (e.g., solutions of where
estimates have a specific value), but as was said, this is not the job of
Kmeans, but rather of general data analysis.

The ellipsoid solution proposed to you is, in a sense, a middle ground
between these two solutions (the circles and the density plots).
You could adjust the (4 or 5) parameters of an ellipsoid to cover "most" of
the points for a particular class and tolerate that the ellipsoids don't
cover a few outliers or artifacts (e.g., the coordinate (3,10) I mentioned
above).
The resulting functional forms of the ellipses might be more precise than
circles and less complex than density contours, and might lead to
actionable knowledge depending on your context/domain.

Hope this helps.
J.B. Brown

2020?12?9?(?) 21:08 Mahmood Naderan <mahmood.nt at gmail.com>:

> >Mebbe principal components analysis would suggest an
> >ellipsoid containing "most" of the points in a "cloud".
>
> Sorry I didn't understand. Can you explain more?
> Regards,
> Mahmood
>
>
>
>
> On Wed, Dec 9, 2020 at 8:55 PM The Helmbolds via scikit-learn <
> scikit-learn at python.org> wrote:
>
>> [scikit-learn] Drawing contours in KMeans4
>>
>>
>> Mebbe principal components analysis would suggest an ellipsoid containing
>> "most" of the points in a "cloud".
>>
>>
>>
>>
>> "You won't find the right answers if you don't ask the right questions!"
>> (Robert Helmbold, 2013)
>>
>>
>> On Wednesday, December 9, 2020, 12:22:49 PM MST, Andrew Howe <
>> ahowe42 at gmail.com> wrote:
>>
>>
>> Ok, I see. Well the attached notebook demonstrates doing this by simply
>> finding the maximum distance from each centroid to it's datapoints and
>> drawing a circle using that radius. It's simple, but will hopefully at
>> least point you in a useful direction.
>> [image: image.png]
>> Andrew
>>
>> <~~~~~~~~~~~~~~~~~~~~~~~~~~~>
>> J. Andrew Howe, PhD
>> LinkedIn Profile <http://www.linkedin.com/in/ahowe42>
>> ResearchGate Profile <http://www.researchgate.net/profile/John_Howe12/>
>> Open Researcher and Contributor ID (ORCID)
>> <http://orcid.org/0000-0002-3553-1990>
>> Github Profile <http://github.com/ahowe42>
>> Personal Website <http://www.andrewhowe.com>
>> I live to learn, so I can learn to live. - me
>> <~~~~~~~~~~~~~~~~~~~~~~~~~~~>
>>
>>
>> On Wed, Dec 9, 2020 at 12:59 PM Mahmood Naderan <mahmood.nt at gmail.com>
>> wrote:
>>
>> I mean a circle/contour to group the points in a cluster for better
>> representation.
>> For example, if there are 6 six clusters, it will be more meaningful to
>> group large data points in a circle or contour.
>>
>> Regards,
>> Mahmood
>>
>>
>>
>>
>> On Wed, Dec 9, 2020 at 11:49 AM Andrew Howe <ahowe42 at gmail.com> wrote:
>>
>> Contours generally indicate a third variable - often a probability
>> density. Kmeans doesn't provide density estimates, so what precisely would
>> you want the contours to represent?
>>
>> Andrew
>>
>> <~~~~~~~~~~~~~~~~~~~~~~~~~~~>
>> J. Andrew Howe, PhD
>> LinkedIn Profile <http://www.linkedin.com/in/ahowe42>
>> ResearchGate Profile <http://www.researchgate.net/profile/John_Howe12/>
>> Open Researcher and Contributor ID (ORCID)
>> <http://orcid.org/0000-0002-3553-1990>
>> Github Profile <http://github.com/ahowe42>
>> Personal Website <http://www.andrewhowe.com>
>> I live to learn, so I can learn to live. - me
>> <~~~~~~~~~~~~~~~~~~~~~~~~~~~>
>>
>>
>> On Wed, Dec 9, 2020 at 9:41 AM Mahmood Naderan <mahmood.nt at gmail.com>
>> wrote:
>>
>> Hi
>> I use the following code to highlight the cluster centers with some red
>> dots.
>>
>> kmeans = KMeans(n_clusters=6, init='k-means++', max_iter=100, n_init=10,
>> random_state=0)
>> pred_y = kmeans.fit_predict(a)
>> plt.scatter(a[:,0], a[:,1])
>> plt.scatter(kmeans.cluster_centers_[:, 0], kmeans.cluster_centers_[:, 1],
>> s=100, c='red')
>> plt.show()
>>
>> I would like to know if it is possible to draw contours over the
>> clusters. Is there any way for that?
>> Please let me know if there is a function or option in KMeans.
>>
>> Regards,
>> Mahmood
>>
>>
>> _______________________________________________
>> scikit-learn mailing list
>> scikit-learn at python.org
>> https://mail.python.org/mailman/listinfo/scikit-learn
>>
>> _______________________________________________
>> scikit-learn mailing list
>> scikit-learn at python.org
>> https://mail.python.org/mailman/listinfo/scikit-learn
>>
>> _______________________________________________
>> scikit-learn mailing list
>> scikit-learn at python.org
>> https://mail.python.org/mailman/listinfo/scikit-learn
>>
>> _______________________________________________
>> scikit-learn mailing list
>> scikit-learn at python.org
>> https://mail.python.org/mailman/listinfo/scikit-learn
>> _______________________________________________
>> scikit-learn mailing list
>> scikit-learn at python.org
>> https://mail.python.org/mailman/listinfo/scikit-learn
>>
> _______________________________________________
> scikit-learn mailing list
> scikit-learn at python.org
> https://mail.python.org/mailman/listinfo/scikit-learn
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mail.python.org/pipermail/scikit-learn/attachments/20201209/2c484e0b/attachment-0001.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: image.png
Type: image/png
Size: 44525 bytes
Desc: not available
URL: <https://mail.python.org/pipermail/scikit-learn/attachments/20201209/2c484e0b/attachment-0001.png>

From abhishek.ghose.82 at gmail.com  Wed Dec  9 16:21:27 2020
From: abhishek.ghose.82 at gmail.com (Abhishek Ghose)
Date: Wed, 9 Dec 2020 13:21:27 -0800
Subject: [scikit-learn] Drawing contours in KMeans
In-Reply-To: <CAJe_vxBnwB_iR4PbX-ADX4DS1oGc4E=VcJS=6mvVB8f5t6H-gA@mail.gmail.com>
References: <CADa2P2VP0sDWyizfEoGm+qfT9dssQoDn2CMGA+xfp+16nnmbUQ@mail.gmail.com>
 <CANnYi3Q+nsF2NzeWPxTSqn+ugRbPJktRMjeLhuGds=Tr0c+kRQ@mail.gmail.com>
 <CADa2P2X1jvzwAWwtGJoZoe2=gq1JPgCj6n7FNNtG-EZB+qmpcg@mail.gmail.com>
 <CANnYi3RvfB6mYxxmXw2XqagjW6dXLQdWnebWYSsVm3OcBkwQgg@mail.gmail.com>
 <694862547.3937714.1607543587026@mail.yahoo.com>
 <CADa2P2WbF=dVz3dQjtMzqEfutC9KrGOm9z_iOJrJXZQ1Sf7sMg@mail.gmail.com>
 <CAJe_vxBnwB_iR4PbX-ADX4DS1oGc4E=VcJS=6mvVB8f5t6H-gA@mail.gmail.com>
Message-ID: <CAOKQjMXObpS-nJr=0vDa8u=VUTDNpUoL_s_Mae4U22K9x5-sGA@mail.gmail.com>

Hi,

A quick way I use is to draw a convex hull (scipy) around the points in a
cluster.
Here's a short example - k-means with k=2 is run on synthetic data:

from sklearn.datasets import make_blobs
from sklearn.cluster import KMeans
from matplotlib import pyplot as plt
from scipy.spatial import ConvexHull


X, _ = make_blobs(centers=2)
kmeans = KMeans(n_clusters=2, random_state=0).fit(X)

# uncomment the next line if you're using a notebook
#%matplotlib inline
for label in set(kmeans.labels_):
    X_clust = X[kmeans.labels_==label]
    hull = ConvexHull(X_clust, qhull_options='QJ')
    vertices_cycle = hull.vertices.tolist()
    vertices_cycle.append(hull.vertices[0])
    plt.plot(X_clust[vertices_cycle, 0], X_clust[vertices_cycle, 1], 'k--',
lw=1)
    plt.scatter(X_clust[:, 0], X_clust[:, 1])

Note:

   1. You can still have overlaps between boundaries - but I think this is
   a good effort-to-results tradeoff.
   2. To draw a closed boundary, you'd need to add the first vertex to the
   list returned by the hull function - the above code does that.
   3. You'd need to handle the case for clusters with <=2 points explicitly
   - not shown in the above code.
   4. I use the "QJ" option (other options at the qhull library page, which
   scipy internally uses: http://www.qhull.org/html/qh-optq.htm) to joggle
   the points a bit when they lie on a line.

Regards


On Wed, Dec 9, 2020 at 12:41 PM Brown J.B. via scikit-learn <
scikit-learn at python.org> wrote:

> Dear Mahmood,
>
> Andrew's solution with a circle will guarantee you render an image in
> which every point is covered within some circle.
>
> However, if data contains outliers or artifacts, you might get circles
> which are excessively large and distort the image you want.
> For example, imagine if there were a single red point in Andrew's image at
> the coordinate (3,10); then, the resulting circle would cover all points in
> the entire plot, which is unlikely what you want.
> You could potentially generate a density estimate for each class and then
> have matplotlib render the contour lines (e.g., solutions of where
> estimates have a specific value), but as was said, this is not the job of
> Kmeans, but rather of general data analysis.
>
> The ellipsoid solution proposed to you is, in a sense, a middle ground
> between these two solutions (the circles and the density plots).
> You could adjust the (4 or 5) parameters of an ellipsoid to cover "most"
> of the points for a particular class and tolerate that the ellipsoids don't
> cover a few outliers or artifacts (e.g., the coordinate (3,10) I mentioned
> above).
> The resulting functional forms of the ellipses might be more precise than
> circles and less complex than density contours, and might lead to
> actionable knowledge depending on your context/domain.
>
> Hope this helps.
> J.B. Brown
>
> 2020?12?9?(?) 21:08 Mahmood Naderan <mahmood.nt at gmail.com>:
>
>> >Mebbe principal components analysis would suggest an
>> >ellipsoid containing "most" of the points in a "cloud".
>>
>> Sorry I didn't understand. Can you explain more?
>> Regards,
>> Mahmood
>>
>>
>>
>>
>> On Wed, Dec 9, 2020 at 8:55 PM The Helmbolds via scikit-learn <
>> scikit-learn at python.org> wrote:
>>
>>> [scikit-learn] Drawing contours in KMeans4
>>>
>>>
>>> Mebbe principal components analysis would suggest an ellipsoid
>>> containing "most" of the points in a "cloud".
>>>
>>>
>>>
>>>
>>> "You won't find the right answers if you don't ask the right questions!"
>>> (Robert Helmbold, 2013)
>>>
>>>
>>> On Wednesday, December 9, 2020, 12:22:49 PM MST, Andrew Howe <
>>> ahowe42 at gmail.com> wrote:
>>>
>>>
>>> Ok, I see. Well the attached notebook demonstrates doing this by simply
>>> finding the maximum distance from each centroid to it's datapoints and
>>> drawing a circle using that radius. It's simple, but will hopefully at
>>> least point you in a useful direction.
>>> [image: image.png]
>>> Andrew
>>>
>>> <~~~~~~~~~~~~~~~~~~~~~~~~~~~>
>>> J. Andrew Howe, PhD
>>> LinkedIn Profile <http://www.linkedin.com/in/ahowe42>
>>> ResearchGate Profile <http://www.researchgate.net/profile/John_Howe12/>
>>> Open Researcher and Contributor ID (ORCID)
>>> <http://orcid.org/0000-0002-3553-1990>
>>> Github Profile <http://github.com/ahowe42>
>>> Personal Website <http://www.andrewhowe.com>
>>> I live to learn, so I can learn to live. - me
>>> <~~~~~~~~~~~~~~~~~~~~~~~~~~~>
>>>
>>>
>>> On Wed, Dec 9, 2020 at 12:59 PM Mahmood Naderan <mahmood.nt at gmail.com>
>>> wrote:
>>>
>>> I mean a circle/contour to group the points in a cluster for better
>>> representation.
>>> For example, if there are 6 six clusters, it will be more meaningful to
>>> group large data points in a circle or contour.
>>>
>>> Regards,
>>> Mahmood
>>>
>>>
>>>
>>>
>>> On Wed, Dec 9, 2020 at 11:49 AM Andrew Howe <ahowe42 at gmail.com> wrote:
>>>
>>> Contours generally indicate a third variable - often a probability
>>> density. Kmeans doesn't provide density estimates, so what precisely would
>>> you want the contours to represent?
>>>
>>> Andrew
>>>
>>> <~~~~~~~~~~~~~~~~~~~~~~~~~~~>
>>> J. Andrew Howe, PhD
>>> LinkedIn Profile <http://www.linkedin.com/in/ahowe42>
>>> ResearchGate Profile <http://www.researchgate.net/profile/John_Howe12/>
>>> Open Researcher and Contributor ID (ORCID)
>>> <http://orcid.org/0000-0002-3553-1990>
>>> Github Profile <http://github.com/ahowe42>
>>> Personal Website <http://www.andrewhowe.com>
>>> I live to learn, so I can learn to live. - me
>>> <~~~~~~~~~~~~~~~~~~~~~~~~~~~>
>>>
>>>
>>> On Wed, Dec 9, 2020 at 9:41 AM Mahmood Naderan <mahmood.nt at gmail.com>
>>> wrote:
>>>
>>> Hi
>>> I use the following code to highlight the cluster centers with some red
>>> dots.
>>>
>>> kmeans = KMeans(n_clusters=6, init='k-means++', max_iter=100, n_init=10,
>>> random_state=0)
>>> pred_y = kmeans.fit_predict(a)
>>> plt.scatter(a[:,0], a[:,1])
>>> plt.scatter(kmeans.cluster_centers_[:, 0], kmeans.cluster_centers_[:,
>>> 1], s=100, c='red')
>>> plt.show()
>>>
>>> I would like to know if it is possible to draw contours over the
>>> clusters. Is there any way for that?
>>> Please let me know if there is a function or option in KMeans.
>>>
>>> Regards,
>>> Mahmood
>>>
>>>
>>> _______________________________________________
>>> scikit-learn mailing list
>>> scikit-learn at python.org
>>> https://mail.python.org/mailman/listinfo/scikit-learn
>>>
>>> _______________________________________________
>>> scikit-learn mailing list
>>> scikit-learn at python.org
>>> https://mail.python.org/mailman/listinfo/scikit-learn
>>>
>>> _______________________________________________
>>> scikit-learn mailing list
>>> scikit-learn at python.org
>>> https://mail.python.org/mailman/listinfo/scikit-learn
>>>
>>> _______________________________________________
>>> scikit-learn mailing list
>>> scikit-learn at python.org
>>> https://mail.python.org/mailman/listinfo/scikit-learn
>>> _______________________________________________
>>> scikit-learn mailing list
>>> scikit-learn at python.org
>>> https://mail.python.org/mailman/listinfo/scikit-learn
>>>
>> _______________________________________________
>> scikit-learn mailing list
>> scikit-learn at python.org
>> https://mail.python.org/mailman/listinfo/scikit-learn
>>
> _______________________________________________
> scikit-learn mailing list
> scikit-learn at python.org
> https://mail.python.org/mailman/listinfo/scikit-learn
>


-- 
Computers: The eventual realization of Douglas Adams' musings - the world
depends on machines controlled by mice.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mail.python.org/pipermail/scikit-learn/attachments/20201209/09ad1340/attachment-0001.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: image.png
Type: image/png
Size: 44525 bytes
Desc: not available
URL: <https://mail.python.org/pipermail/scikit-learn/attachments/20201209/09ad1340/attachment-0001.png>

From abhishek.ghose.82 at gmail.com  Wed Dec  9 19:25:38 2020
From: abhishek.ghose.82 at gmail.com (Abhishek Ghose)
Date: Wed, 9 Dec 2020 16:25:38 -0800
Subject: [scikit-learn] Drawing contours in KMeans
In-Reply-To: <CAOKQjMXObpS-nJr=0vDa8u=VUTDNpUoL_s_Mae4U22K9x5-sGA@mail.gmail.com>
References: <CADa2P2VP0sDWyizfEoGm+qfT9dssQoDn2CMGA+xfp+16nnmbUQ@mail.gmail.com>
 <CANnYi3Q+nsF2NzeWPxTSqn+ugRbPJktRMjeLhuGds=Tr0c+kRQ@mail.gmail.com>
 <CADa2P2X1jvzwAWwtGJoZoe2=gq1JPgCj6n7FNNtG-EZB+qmpcg@mail.gmail.com>
 <CANnYi3RvfB6mYxxmXw2XqagjW6dXLQdWnebWYSsVm3OcBkwQgg@mail.gmail.com>
 <694862547.3937714.1607543587026@mail.yahoo.com>
 <CADa2P2WbF=dVz3dQjtMzqEfutC9KrGOm9z_iOJrJXZQ1Sf7sMg@mail.gmail.com>
 <CAJe_vxBnwB_iR4PbX-ADX4DS1oGc4E=VcJS=6mvVB8f5t6H-gA@mail.gmail.com>
 <CAOKQjMXObpS-nJr=0vDa8u=VUTDNpUoL_s_Mae4U22K9x5-sGA@mail.gmail.com>
Message-ID: <CAOKQjMVnK=A483Pr2ePz_2JxB+FjDmSOc869cGgNAyGBk4rj0A@mail.gmail.com>

Sorry, just noticed that I had forgotten to attach a sample image.

Regards

On Wed, Dec 9, 2020 at 1:21 PM Abhishek Ghose <abhishek.ghose.82 at gmail.com>
wrote:

> Hi,
>
> A quick way I use is to draw a convex hull (scipy) around the points in a
> cluster.
> Here's a short example - k-means with k=2 is run on synthetic data:
>
> from sklearn.datasets import make_blobs
> from sklearn.cluster import KMeans
> from matplotlib import pyplot as plt
> from scipy.spatial import ConvexHull
>
>
> X, _ = make_blobs(centers=2)
> kmeans = KMeans(n_clusters=2, random_state=0).fit(X)
>
> # uncomment the next line if you're using a notebook
> #%matplotlib inline
> for label in set(kmeans.labels_):
>     X_clust = X[kmeans.labels_==label]
>     hull = ConvexHull(X_clust, qhull_options='QJ')
>     vertices_cycle = hull.vertices.tolist()
>     vertices_cycle.append(hull.vertices[0])
>     plt.plot(X_clust[vertices_cycle, 0], X_clust[vertices_cycle, 1],
> 'k--', lw=1)
>     plt.scatter(X_clust[:, 0], X_clust[:, 1])
>
> Note:
>
>    1. You can still have overlaps between boundaries - but I think this
>    is a good effort-to-results tradeoff.
>    2. To draw a closed boundary, you'd need to add the first vertex to
>    the list returned by the hull function - the above code does that.
>    3. You'd need to handle the case for clusters with <=2 points
>    explicitly - not shown in the above code.
>    4. I use the "QJ" option (other options at the qhull library page,
>    which scipy internally uses: http://www.qhull.org/html/qh-optq.htm) to
>    joggle the points a bit when they lie on a line.
>
> Regards
>
>
> On Wed, Dec 9, 2020 at 12:41 PM Brown J.B. via scikit-learn <
> scikit-learn at python.org> wrote:
>
>> Dear Mahmood,
>>
>> Andrew's solution with a circle will guarantee you render an image in
>> which every point is covered within some circle.
>>
>> However, if data contains outliers or artifacts, you might get circles
>> which are excessively large and distort the image you want.
>> For example, imagine if there were a single red point in Andrew's image
>> at the coordinate (3,10); then, the resulting circle would cover all points
>> in the entire plot, which is unlikely what you want.
>> You could potentially generate a density estimate for each class and then
>> have matplotlib render the contour lines (e.g., solutions of where
>> estimates have a specific value), but as was said, this is not the job of
>> Kmeans, but rather of general data analysis.
>>
>> The ellipsoid solution proposed to you is, in a sense, a middle ground
>> between these two solutions (the circles and the density plots).
>> You could adjust the (4 or 5) parameters of an ellipsoid to cover "most"
>> of the points for a particular class and tolerate that the ellipsoids don't
>> cover a few outliers or artifacts (e.g., the coordinate (3,10) I mentioned
>> above).
>> The resulting functional forms of the ellipses might be more precise than
>> circles and less complex than density contours, and might lead to
>> actionable knowledge depending on your context/domain.
>>
>> Hope this helps.
>> J.B. Brown
>>
>> 2020?12?9?(?) 21:08 Mahmood Naderan <mahmood.nt at gmail.com>:
>>
>>> >Mebbe principal components analysis would suggest an
>>> >ellipsoid containing "most" of the points in a "cloud".
>>>
>>> Sorry I didn't understand. Can you explain more?
>>> Regards,
>>> Mahmood
>>>
>>>
>>>
>>>
>>> On Wed, Dec 9, 2020 at 8:55 PM The Helmbolds via scikit-learn <
>>> scikit-learn at python.org> wrote:
>>>
>>>> [scikit-learn] Drawing contours in KMeans4
>>>>
>>>>
>>>> Mebbe principal components analysis would suggest an ellipsoid
>>>> containing "most" of the points in a "cloud".
>>>>
>>>>
>>>>
>>>>
>>>> "You won't find the right answers if you don't ask the right
>>>> questions!" (Robert Helmbold, 2013)
>>>>
>>>>
>>>> On Wednesday, December 9, 2020, 12:22:49 PM MST, Andrew Howe <
>>>> ahowe42 at gmail.com> wrote:
>>>>
>>>>
>>>> Ok, I see. Well the attached notebook demonstrates doing this by simply
>>>> finding the maximum distance from each centroid to it's datapoints and
>>>> drawing a circle using that radius. It's simple, but will hopefully at
>>>> least point you in a useful direction.
>>>> [image: image.png]
>>>> Andrew
>>>>
>>>> <~~~~~~~~~~~~~~~~~~~~~~~~~~~>
>>>> J. Andrew Howe, PhD
>>>> LinkedIn Profile <http://www.linkedin.com/in/ahowe42>
>>>> ResearchGate Profile <http://www.researchgate.net/profile/John_Howe12/>
>>>> Open Researcher and Contributor ID (ORCID)
>>>> <http://orcid.org/0000-0002-3553-1990>
>>>> Github Profile <http://github.com/ahowe42>
>>>> Personal Website <http://www.andrewhowe.com>
>>>> I live to learn, so I can learn to live. - me
>>>> <~~~~~~~~~~~~~~~~~~~~~~~~~~~>
>>>>
>>>>
>>>> On Wed, Dec 9, 2020 at 12:59 PM Mahmood Naderan <mahmood.nt at gmail.com>
>>>> wrote:
>>>>
>>>> I mean a circle/contour to group the points in a cluster for better
>>>> representation.
>>>> For example, if there are 6 six clusters, it will be more meaningful to
>>>> group large data points in a circle or contour.
>>>>
>>>> Regards,
>>>> Mahmood
>>>>
>>>>
>>>>
>>>>
>>>> On Wed, Dec 9, 2020 at 11:49 AM Andrew Howe <ahowe42 at gmail.com> wrote:
>>>>
>>>> Contours generally indicate a third variable - often a probability
>>>> density. Kmeans doesn't provide density estimates, so what precisely would
>>>> you want the contours to represent?
>>>>
>>>> Andrew
>>>>
>>>> <~~~~~~~~~~~~~~~~~~~~~~~~~~~>
>>>> J. Andrew Howe, PhD
>>>> LinkedIn Profile <http://www.linkedin.com/in/ahowe42>
>>>> ResearchGate Profile <http://www.researchgate.net/profile/John_Howe12/>
>>>> Open Researcher and Contributor ID (ORCID)
>>>> <http://orcid.org/0000-0002-3553-1990>
>>>> Github Profile <http://github.com/ahowe42>
>>>> Personal Website <http://www.andrewhowe.com>
>>>> I live to learn, so I can learn to live. - me
>>>> <~~~~~~~~~~~~~~~~~~~~~~~~~~~>
>>>>
>>>>
>>>> On Wed, Dec 9, 2020 at 9:41 AM Mahmood Naderan <mahmood.nt at gmail.com>
>>>> wrote:
>>>>
>>>> Hi
>>>> I use the following code to highlight the cluster centers with some red
>>>> dots.
>>>>
>>>> kmeans = KMeans(n_clusters=6, init='k-means++', max_iter=100,
>>>> n_init=10, random_state=0)
>>>> pred_y = kmeans.fit_predict(a)
>>>> plt.scatter(a[:,0], a[:,1])
>>>> plt.scatter(kmeans.cluster_centers_[:, 0], kmeans.cluster_centers_[:,
>>>> 1], s=100, c='red')
>>>> plt.show()
>>>>
>>>> I would like to know if it is possible to draw contours over the
>>>> clusters. Is there any way for that?
>>>> Please let me know if there is a function or option in KMeans.
>>>>
>>>> Regards,
>>>> Mahmood
>>>>
>>>>
>>>> _______________________________________________
>>>> scikit-learn mailing list
>>>> scikit-learn at python.org
>>>> https://mail.python.org/mailman/listinfo/scikit-learn
>>>>
>>>> _______________________________________________
>>>> scikit-learn mailing list
>>>> scikit-learn at python.org
>>>> https://mail.python.org/mailman/listinfo/scikit-learn
>>>>
>>>> _______________________________________________
>>>> scikit-learn mailing list
>>>> scikit-learn at python.org
>>>> https://mail.python.org/mailman/listinfo/scikit-learn
>>>>
>>>> _______________________________________________
>>>> scikit-learn mailing list
>>>> scikit-learn at python.org
>>>> https://mail.python.org/mailman/listinfo/scikit-learn
>>>> _______________________________________________
>>>> scikit-learn mailing list
>>>> scikit-learn at python.org
>>>> https://mail.python.org/mailman/listinfo/scikit-learn
>>>>
>>> _______________________________________________
>>> scikit-learn mailing list
>>> scikit-learn at python.org
>>> https://mail.python.org/mailman/listinfo/scikit-learn
>>>
>> _______________________________________________
>> scikit-learn mailing list
>> scikit-learn at python.org
>> https://mail.python.org/mailman/listinfo/scikit-learn
>>
>
>
> --
> Computers: The eventual realization of Douglas Adams' musings - the world
> depends on machines controlled by mice.
>


-- 
Computers: The eventual realization of Douglas Adams' musings - the world
depends on machines controlled by mice.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mail.python.org/pipermail/scikit-learn/attachments/20201209/92e35350/attachment-0001.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: image.png
Type: image/png
Size: 44525 bytes
Desc: not available
URL: <https://mail.python.org/pipermail/scikit-learn/attachments/20201209/92e35350/attachment-0002.png>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: kmeans_convexhull.PNG
Type: image/png
Size: 29519 bytes
Desc: not available
URL: <https://mail.python.org/pipermail/scikit-learn/attachments/20201209/92e35350/attachment-0003.png>

From mahmood.nt at gmail.com  Mon Dec 14 11:31:24 2020
From: mahmood.nt at gmail.com (Mahmood Naderan)
Date: Mon, 14 Dec 2020 17:31:24 +0100
Subject: [scikit-learn] Using text labels in dendrogram
Message-ID: <CADa2P2Xa-z_S-fHBrwUt8tECq5-2E6_Vmk0dR6ahcsHYAjiUEQ@mail.gmail.com>

Hi
I use the following code to create a dendrogram from a set of x-y points

import matplotlib.pyplot as plt
import numpy as np
from scipy.cluster.hierarchy import dendrogram, linkage
a = np.array([
[5.840,-2.339],
[6.320,-2.665],
[-1.698,-0.084],
],)
linked = linkage(a, 'single')
labelList = range(1, 69)
dendrogram(linked,
            orientation='top',
            labels=labelList,
            distance_sort='descending',
            show_leaf_counts=True)
plt.show()

This code automatically assigns labels for the points and the result is not
sorted. For example, the x-axis labels are 2 1 3.
I want to assign a text for each point. Then no matter where the point
goes, I can see the text that I have assigned. For example, I want to see
PDD, XDD, BDD.
How can I create a list of text labels and bind that to the array a?

Regards,
Mahmood
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mail.python.org/pipermail/scikit-learn/attachments/20201214/1cfd74db/attachment.html>

From mahmood.nt at gmail.com  Tue Dec 15 01:36:57 2020
From: mahmood.nt at gmail.com (Mahmood Naderan)
Date: Tue, 15 Dec 2020 07:36:57 +0100
Subject: [scikit-learn] Using text labels in dendrogram
In-Reply-To: <CADa2P2Xa-z_S-fHBrwUt8tECq5-2E6_Vmk0dR6ahcsHYAjiUEQ@mail.gmail.com>
References: <CADa2P2Xa-z_S-fHBrwUt8tECq5-2E6_Vmk0dR6ahcsHYAjiUEQ@mail.gmail.com>
Message-ID: <CADa2P2UL4XTppu0o5--WY33uHgsR732uFYq1wgBf5Z=bfLE92A@mail.gmail.com>

OK. As I checked the documents, I have to fix the labelList line to the
string that I want.

Regards,
Mahmood




On Mon, Dec 14, 2020 at 5:31 PM Mahmood Naderan <mahmood.nt at gmail.com>
wrote:

> Hi
> I use the following code to create a dendrogram from a set of x-y points
>
> import matplotlib.pyplot as plt
> import numpy as np
> from scipy.cluster.hierarchy import dendrogram, linkage
> a = np.array([
> [5.840,-2.339],
> [6.320,-2.665],
> [-1.698,-0.084],
> ],)
> linked = linkage(a, 'single')
> labelList = range(1, 69)
> dendrogram(linked,
>             orientation='top',
>             labels=labelList,
>             distance_sort='descending',
>             show_leaf_counts=True)
> plt.show()
>
> This code automatically assigns labels for the points and the result is
> not sorted. For example, the x-axis labels are 2 1 3.
> I want to assign a text for each point. Then no matter where the point
> goes, I can see the text that I have assigned. For example, I want to see
> PDD, XDD, BDD.
> How can I create a list of text labels and bind that to the array a?
>
> Regards,
> Mahmood
>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mail.python.org/pipermail/scikit-learn/attachments/20201215/8593874b/attachment.html>

From adrin.jalali at gmail.com  Thu Dec 17 02:10:04 2020
From: adrin.jalali at gmail.com (Adrin)
Date: Thu, 17 Dec 2020 08:10:04 +0100
Subject: [scikit-learn] Daily failure: Run failed: Wheel builder - master
 (1e53ae2)
In-Reply-To: <adrinjalali/scikit-learn/check-suites/MDEwOkNoZWNrU3VpdGUxNjgzMjUzMDE1/1608178859@github.com>
References: <adrinjalali/scikit-learn/check-suites/MDEwOkNoZWNrU3VpdGUxNjgzMjUzMDE1/1608178859@github.com>
Message-ID: <CAEOrW4-88pNWs_F3B0PerJHU+z5SSky3yP4bQRQ0cdCsa9Q+gA@mail.gmail.com>

Hi,

I keep getting this email every night, sending it here to make sure we're
aware of it.

Cheers,
Adrin


---------- Forwarded message ---------
From: Adrin Jalali <notifications at github.com>
Date: Thu., Dec. 17, 2020, 05:21
Subject: [adrinjalali/scikit-learn] Run failed: Wheel builder - master
(1e53ae2)
To: adrinjalali/scikit-learn <scikit-learn at noreply.github.com>
Cc: Ci activity <ci_activity at noreply.github.com>



[image: GitHub] [adrinjalali/scikit-learn] Wheel builder workflow run

  Wheel builder: Some jobs were not successful

View workflow run
<https://github.com/adrinjalali/scikit-learn/actions/runs/427220167>

[image: Check build trigger]

*Wheel builder* / Check build trigger
Succeeded in 17 seconds
[image: Build wheel for cp36-win32]

*Wheel builder* / Build wheel for cp36-win32
Succeeded in 21 minutes
[image: Build wheel for cp36-win_amd64]

*Wheel builder* / Build wheel for cp36-win_amd64
Succeeded in 20 minutes and 3 seconds
[image: Build wheel for cp37-win32]

*Wheel builder* / Build wheel for cp37-win32
Succeeded in 19 minutes and 20 seconds
[image: Build wheel for cp37-win_amd64]

*Wheel builder* / Build wheel for cp37-win_amd64
Succeeded in 22 minutes and 28 seconds
[image: Build wheel for cp38-win32]

*Wheel builder* / Build wheel for cp38-win32
Succeeded in 19 minutes and 39 seconds
[image: Build wheel for cp38-win_amd64]

*Wheel builder* / Build wheel for cp38-win_amd64
Succeeded in 21 minutes and 21 seconds
[image: Build wheel for cp39-win32]

*Wheel builder* / Build wheel for cp39-win32
Succeeded in 20 minutes and 37 seconds
[image: Build wheel for cp39-win_amd64]

*Wheel builder* / Build wheel for cp39-win_amd64
Succeeded in 24 minutes and 7 seconds
[image: Build wheel for cp36-manylinux_i686]

*Wheel builder* / Build wheel for cp36-manylinux_i686
Succeeded in 17 minutes and 50 seconds
View all 23 job statuses
<https://github.com/adrinjalali/scikit-learn/actions/runs/427220167>




?
You are receiving this because this workflow ran on your branch.
Manage your GitHub Actions notifications
<https://github.com/settings/notifications>


GitHub, Inc. ?88 Colin P Kelly Jr Street ?San Francisco, CA 94107
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mail.python.org/pipermail/scikit-learn/attachments/20201217/0c0290e3/attachment-0001.html>

From loic.esteve at ymail.com  Thu Dec 17 02:42:36 2020
From: loic.esteve at ymail.com (=?utf-8?B?TG/Dr2MgRXN0w6h2ZQ==?=)
Date: Thu, 17 Dec 2020 08:42:36 +0100
Subject: [scikit-learn] Daily failure: Run failed: Wheel builder -
 master (1e53ae2)
In-Reply-To: <CAEOrW4-88pNWs_F3B0PerJHU+z5SSky3yP4bQRQ0cdCsa9Q+gA@mail.gmail.com>
References: <adrinjalali/scikit-learn/check-suites/MDEwOkNoZWNrU3VpdGUxNjgzMjUzMDE1/1608178859@github.com>
 <CAEOrW4-88pNWs_F3B0PerJHU+z5SSky3yP4bQRQ0cdCsa9Q+gA@mail.gmail.com>
Message-ID: <vnok360498ur.fsf@ymail.com>

Me too, as a quick fix I reverted my fork (lesteve/scikit-learn) master
branch to an old version to avoid this ...

The better fix would be to modify the Github Action so that it does not
run in forks but only in the main repo i.e. scikit-learn/scikit-learn.

Cheers,
Lo?c

> Hi,
>
> I keep getting this email every night, sending it here to make sure we're aware of it.
>
> Cheers,
> Adrin
>
> ---------- Forwarded message ---------
> From: Adrin Jalali <notifications at github.com>
> Date: Thu., Dec. 17, 2020, 05:21
> Subject: [adrinjalali/scikit-learn] Run failed: Wheel builder - master (1e53ae2)
> To: adrinjalali/scikit-learn <scikit-learn at noreply.github.com>
> Cc: Ci activity <ci_activity at noreply.github.com>
>
>     
> GitHub   
>    
> [adrinjalali/scikit-learn] Wheel builder workflow run    
>  
> *      
>       
>         
> Wheel builder: Some jobs were not successful      
>       
>         
> View workflow run         
>     
>  Check build trigger Wheel builder / Check build trigger        
>   Succeeded in 17 seconds        
>     
>  Build wheel for cp36-win32 Wheel builder / Build wheel for cp36-win32        
>   Succeeded in 21 minutes        
>     
>  Build wheel for cp36-win_amd64 Wheel builder / Build wheel for cp36-win_amd64        
>   Succeeded in 20 minutes and 3 seconds        
>     
>  Build wheel for cp37-win32 Wheel builder / Build wheel for cp37-win32        
>   Succeeded in 19 minutes and 20 seconds        
>     
>  Build wheel for cp37-win_amd64 Wheel builder / Build wheel for cp37-win_amd64        
>   Succeeded in 22 minutes and 28 seconds        
>     
>  Build wheel for cp38-win32 Wheel builder / Build wheel for cp38-win32        
>   Succeeded in 19 minutes and 39 seconds        
>     
>  Build wheel for cp38-win_amd64 Wheel builder / Build wheel for cp38-win_amd64        
>   Succeeded in 21 minutes and 21 seconds        
>     
>  Build wheel for cp39-win32 Wheel builder / Build wheel for cp39-win32        
>   Succeeded in 20 minutes and 37 seconds        
>     
>  Build wheel for cp39-win_amd64 Wheel builder / Build wheel for cp39-win_amd64        
>   Succeeded in 24 minutes and 7 seconds        
>     
>  Build wheel for cp36-manylinux_i686 Wheel builder / Build wheel for cp36-manylinux_i686        
>   Succeeded in 17 minutes and 50 seconds        
>     
> View all 23 job statuses        
>  
>     
>     
> ?  
> You are receiving this because this workflow ran on your branch.  
> Manage your GitHub Actions notifications  
>  
>     
> GitHub, Inc. ?88 Colin P Kelly Jr Street ?San Francisco, CA 94107  
> GitHub
> *
> Check build trigger
> Build wheel for cp36-win32
> Build wheel for cp36-win_amd64
> Build wheel for cp37-win32
> Build wheel for cp37-win_amd64
> Build wheel for cp38-win32
> Build wheel for cp38-win_amd64
> Build wheel for cp39-win32
> Build wheel for cp39-win_amd64
> Build wheel for cp36-manylinux_i686


From g.lemaitre58 at gmail.com  Tue Dec 22 12:11:05 2020
From: g.lemaitre58 at gmail.com (=?UTF-8?Q?Guillaume_Lema=C3=AEtre?=)
Date: Tue, 22 Dec 2020 18:11:05 +0100
Subject: [scikit-learn] ANN scikit-learn 0.24.0 release
Message-ID: <CACDxx9iQ864BSSudo17MwdD+2c=JEA6vVjB3bTbmFACTOww7=g@mail.gmail.com>

We're happy to announce the 0.24.0 release and already out on PyPI and
conda-forge.

You can read the release highlights under
https://scikit-learn.org/stable/auto_examples/release_highlights/plot_release_highlights_0_24_0.html
and the long version of the change log under
https://scikit-learn.org/stable/whats_new/v0.24.html#version-0-24-0

New major features include: Highlights include successive halving,
categorical support for GBRT,
SelfTrainingClassifier, SequentialFeatureSelector and much more!

This version supports Python versions 3.6 to 3.9. You can give it a go
using `pip install -U scikit-learn` or `conda install -c conda-forge
scikit-learn`.

A big thanks to all contributors for making this release possible.

Regards,
On the behalf of the scikit-learn maintainer team.
-- 
Guillaume Lemaitre
Scikit-learn @ Inria Foundation
https://glemaitre.github.io/
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mail.python.org/pipermail/scikit-learn/attachments/20201222/27184fdb/attachment.html>

From joel.nothman at gmail.com  Tue Dec 22 22:00:44 2020
From: joel.nothman at gmail.com (Joel Nothman)
Date: Wed, 23 Dec 2020 14:00:44 +1100
Subject: [scikit-learn] ANN scikit-learn 0.24.0 release
In-Reply-To: <CACDxx9iQ864BSSudo17MwdD+2c=JEA6vVjB3bTbmFACTOww7=g@mail.gmail.com>
References: <CACDxx9iQ864BSSudo17MwdD+2c=JEA6vVjB3bTbmFACTOww7=g@mail.gmail.com>
Message-ID: <CAAkaFLXwJrzPnkDk36n0nC1CWEK_Z6FwZ1SkFXDDcRuEmGrH3A@mail.gmail.com>

Thanks and congrats to all involved! Some very helpful features. And rumour
has it we might have version 1.0 in 2021... :-o

On Wed, 23 Dec 2020, 4:12 am Guillaume Lema?tre, <g.lemaitre58 at gmail.com>
wrote:

> We're happy to announce the 0.24.0 release and already out on PyPI and
> conda-forge.
>
> You can read the release highlights under
> https://scikit-learn.org/stable/auto_examples/release_highlights/plot_release_highlights_0_24_0.html
> and the long version of the change log under
> https://scikit-learn.org/stable/whats_new/v0.24.html#version-0-24-0
>
> New major features include: Highlights include successive halving,
> categorical support for GBRT,
> SelfTrainingClassifier, SequentialFeatureSelector and much more!
>
> This version supports Python versions 3.6 to 3.9. You can give it a go
> using `pip install -U scikit-learn` or `conda install -c conda-forge
> scikit-learn`.
>
> A big thanks to all contributors for making this release possible.
>
> Regards,
> On the behalf of the scikit-learn maintainer team.
> --
> Guillaume Lemaitre
> Scikit-learn @ Inria Foundation
> https://glemaitre.github.io/
> _______________________________________________
> scikit-learn mailing list
> scikit-learn at python.org
> https://mail.python.org/mailman/listinfo/scikit-learn
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mail.python.org/pipermail/scikit-learn/attachments/20201223/8c352542/attachment.html>

From mahmood.nt at gmail.com  Fri Dec 25 09:16:02 2020
From: mahmood.nt at gmail.com (Mahmood Naderan)
Date: Fri, 25 Dec 2020 15:16:02 +0100
Subject: [scikit-learn] Comparing Scikit and Xlstat for PCA analysis
Message-ID: <CADa2P2VXhRo-gAR_vjbTLq08g8X54Mr1rN_BB=Eh-HdLUwxuew@mail.gmail.com>

Hi
I have a test csv file and I have written a code to show the PCA for that.
I also use another tool in Excel (XLSTAT) to compare the results.
The XLSTAT automatically calculates the number of features, however, based
on my understanding, I have to specify how many components are needed using
the scikit package. For example, while XLSTAT shows 5 features:

Factor scores:
F1 F2 F3 F4 F5
A1 -1.293 -0.663 -0.462 -0.713 0.010
A2 -0.297 0.293 -1.429 0.397 0.056
A3 2.328 0.069 0.987 -0.108 0.062
A4 -0.556 -2.273 0.538 0.344 -0.032
A5 1.823 0.775 -0.597 -0.052 -0.085
A6 -2.005 1.799 0.963 0.133 -0.011


In the following code, I specified 2 components:

x = StandardScaler().fit_transform(x)
pca = PCA(n_components=2)
principalComponents = pca.fit_transform(x)
print( principalComponents )

[[-1.29292842 0.66325508] [-0.29706395 -0.29346337] [ 2.32751305
-0.06850045] [-0.5558091 2.27288988] [ 1.82312052 -0.77527304] [-2.0048321
-1.7989081 ]]



As you can see, the first column in XLSTAT and scikit are the same.
However, the second columns are negated.
For example, considering F1 and F2, we see

XLSTAT => -1.293 -0.663
scikit =>  [-1.29292842 0.66325508]


So, my questions are

1) Isn't there any way to use scikit for an unknown number of principal
components? So that I can query the number of principal components and use
a scree plot then.

2) Considering the F1 and F2 as a XY scatter point, I want to know why the
value of Y in XLSTAT and scikit are opposite?

The code which I write is available at https://pastebin.com/ghJQ6L4C
Any idea?
Regards,
Mahmood
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mail.python.org/pipermail/scikit-learn/attachments/20201225/939b8693/attachment.html>

From davidcee at hotmail.com  Sun Dec 27 21:36:59 2020
From: davidcee at hotmail.com (DAVID cofield)
Date: Mon, 28 Dec 2020 02:36:59 +0000
Subject: [scikit-learn] Interpreting results of random forest classifier
Message-ID: <YQXPR01MB272504EAD60C0F89DD07B2C1D2D90@YQXPR01MB2725.CANPRD01.PROD.OUTLOOK.COM>

Could someone explain what the probability values provided by the random forest classifier represents?

When I run the classifier with two classes, I get prediction values and associated to these prediction values are probabilities. As I understand it now the 0 probability is probability that the prediction is wrong, and the 1 probability is the probability that the prediction is correct. One thing I do not understand is why the probability ranges do not go from 0 to 100; they go from 0 to 49 for 0 probability and 49-100 for 1 probability. How do I interpret  the probabilities?

Sent from Mail<https://go.microsoft.com/fwlink/?LinkId=550986> for Windows 10

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mail.python.org/pipermail/scikit-learn/attachments/20201228/b17f7b48/attachment.html>

From marmochiaskl at gmail.com  Mon Dec 28 06:31:30 2020
From: marmochiaskl at gmail.com (Chiara Marmo)
Date: Mon, 28 Dec 2020 12:31:30 +0100
Subject: [scikit-learn] Monthly meeting January 4th 2021
Message-ID: <CAGfF148--mOdB9wXVS3v5D-QVVbrnktZFmOoLOOaPy==qXVqGg@mail.gmail.com>

Dear list,

The first 2021 scikit-learn monthly meeting will take place on Monday
January 4th at 8PM UTC:
https://www.timeanddate.com/worldclock/meetingdetails.html?year=2021&month=01&day=04&hour=20&min=0&sec=0&p1=179&p2=240&p3=195&p4=224

While these meetings are mainly for core-devs to discuss the current
topics, we are also happy to welcome non-core devs and other project
maintainers.

Feel free to join, using the following link:

https://meet.google.com/xhq-yoga-rtf

If you plan to attend and you would like to discuss something specific
about your contribution please add your name (or github pseudo) in the "
Contributors <https://hackmd.io/qtKt7pTNSXanU-MJOIMxbw#Contributors>"
section, of the public pad:

https://hackmd.io/qtKt7pTNSXanU-MJOIMxbw

Best

Chiara
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mail.python.org/pipermail/scikit-learn/attachments/20201228/dd59e917/attachment.html>

From niourf at gmail.com  Mon Dec 28 07:41:46 2020
From: niourf at gmail.com (Nicolas Hug)
Date: Mon, 28 Dec 2020 12:41:46 +0000
Subject: [scikit-learn] Interpreting results of random forest classifier
In-Reply-To: <YQXPR01MB272504EAD60C0F89DD07B2C1D2D90@YQXPR01MB2725.CANPRD01.PROD.OUTLOOK.COM>
References: <YQXPR01MB272504EAD60C0F89DD07B2C1D2D90@YQXPR01MB2725.CANPRD01.PROD.OUTLOOK.COM>
Message-ID: <6906f0d9-d34a-f751-9975-3f3d8d2f3b80@gmail.com>

Hi David,

 > As I understand it now the 0 probability is probability that the 
prediction is wrong, and the 1 probability is the probability that the 
prediction is correct

No: in binary classification, the `predict_proba` method returns a 
single number in [0, 1] indicating the probability that the sample 
belongs to the positive class (1). In other words, ? (proba <= 0.5? iff 
prediction == 0). The threshold is 0.5 since there are only 2 classes.

 > One thing I do not understand is why the probability ranges do not go 
from 0 to 100; they go from 0 to 49 for 0 probability and 49-100 for 1 
probability

This is a correct observation and it's a direct consequence of the above 
definition.

The way probabilities are computed is briefly described here: 
https://scikit-learn.org/stable/modules/generated/sklearn.ensemble.RandomForestClassifier.html#sklearn.ensemble.RandomForestClassifier.predict_proba

Nicolas

On 12/28/20 2:36 AM, DAVID cofield wrote:
>
> Could someone explain what the probability values provided by the 
> random forest classifier represents?
>
> When I run the classifier with two classes, I get prediction values 
> and associated to these prediction values are probabilities. As I 
> understand it now the 0 probability is probability that the prediction 
> is wrong, and the 1 probability is the probability that the prediction 
> is correct. One thing I do not understand is why the probability 
> ranges do not go from 0 to 100; they go from 0 to 49 for 0 probability 
> and 49-100 for 1 probability. How do I interpret ?the probabilities?
>
> Sent from Mail <https://go.microsoft.com/fwlink/?LinkId=550986> for 
> Windows 10
>
>
> _______________________________________________
> scikit-learn mailing list
> scikit-learn at python.org
> https://mail.python.org/mailman/listinfo/scikit-learn
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mail.python.org/pipermail/scikit-learn/attachments/20201228/89b11f56/attachment.html>

From g.lemaitre58 at gmail.com  Mon Dec 28 10:28:06 2020
From: g.lemaitre58 at gmail.com (=?ISO-8859-1?Q?Guillaume_Lema=EEtre?=)
Date: Mon, 28 Dec 2020 16:28:06 +0100
Subject: [scikit-learn] Comparing Scikit and Xlstat for PCA analysis
In-Reply-To: <CADa2P2VXhRo-gAR_vjbTLq08g8X54Mr1rN_BB=Eh-HdLUwxuew@mail.gmail.com>
Message-ID: <fsjpb2mpafl7d4pkm5io89v0.1609169286168@gmail.com>

n_components set to  'auto' is a strategy that will pick the number of components. The sign of the PC does not matter so much since they are still orthogonal. So change will depend of the solver that should be different in both software. 




Sent from my phone - sorry to be brief and potential misspell.



From mahmood.nt at gmail.com  Mon Dec 28 11:52:19 2020
From: mahmood.nt at gmail.com (Mahmood Naderan)
Date: Mon, 28 Dec 2020 17:52:19 +0100
Subject: [scikit-learn] Comparing Scikit and Xlstat for PCA analysis
In-Reply-To: <fsjpb2mpafl7d4pkm5io89v0.1609169286168@gmail.com>
References: <CADa2P2VXhRo-gAR_vjbTLq08g8X54Mr1rN_BB=Eh-HdLUwxuew@mail.gmail.com>
 <fsjpb2mpafl7d4pkm5io89v0.1609169286168@gmail.com>
Message-ID: <CADa2P2UU5T5yeLjNPsfisPTVmH1eCeQLf2=kPcr_ixbRGwG2-w@mail.gmail.com>

Hi Guillaume,
Thanks for the reply. May I know if I can choose different solvers in the
scikit package or not.

Regards,
Mahmood




On Mon, Dec 28, 2020 at 4:30 PM Guillaume Lema?tre <g.lemaitre58 at gmail.com>
wrote:

> n_components set to  'auto' is a strategy that will pick the number of
> components. The sign of the PC does not matter so much since they are still
> orthogonal. So change will depend of the solver that should be different in
> both software.
>
>
>
>
> Sent from my phone - sorry to be brief and potential misspell.
>
>
> _______________________________________________
> scikit-learn mailing list
> scikit-learn at python.org
> https://mail.python.org/mailman/listinfo/scikit-learn
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mail.python.org/pipermail/scikit-learn/attachments/20201228/44c8fb4c/attachment.html>