2013/2/23 Andreas Mueller :
> I forgot you wanted also do include a new joblib, right?
> That is in master, isn't it? Not sure if Lars already picked it.
Yeah, I think I did. Maybe cherry-pick pprett's
e6c85a36a6dd8a45cb8d7fcd6b8015e65af91e3d (sparse coef_ support for
linear classifiers) as a hidd
and please include fixes for roc_curve (there were two commits). For
now I have picked them up for the (Neuro)Debian package
cheers
On Fri, 22 Feb 2013, Andreas Mueller wrote:
> Hey everybody.
> So I plan to do a bugfix release based on Lars' 0.13 branch tomorrow.
> I also want to includes Yaro
On 02/23/2013 12:12 AM, Andreas Mueller wrote:
> On 02/23/2013 12:05 AM, Gael Varoquaux wrote:
>> On Fri, Feb 22, 2013 at 11:07:13PM +0100, Andreas Mueller wrote:
>>> So I plan to do a bugfix release based on Lars' 0.13 branch tomorrow.
>>> I also want to includes Yaroslavs train_test_split fix.
>>
On 02/23/2013 12:05 AM, Gael Varoquaux wrote:
> On Fri, Feb 22, 2013 at 11:07:13PM +0100, Andreas Mueller wrote:
>> So I plan to do a bugfix release based on Lars' 0.13 branch tomorrow.
>> I also want to includes Yaroslavs train_test_split fix.
> What's your schedule during the day? What's the rema
On Fri, Feb 22, 2013 at 11:07:13PM +0100, Andreas Mueller wrote:
> So I plan to do a bugfix release based on Lars' 0.13 branch tomorrow.
> I also want to includes Yaroslavs train_test_split fix.
What's your schedule during the day? What's the remaining work to do?
I'll try to pitch in.
G
---
Hey everybody.
So I plan to do a bugfix release based on Lars' 0.13 branch tomorrow.
I also want to includes Yaroslavs train_test_split fix.
Anything else?
Cheers,
Andy
--
Everyone hates slow websites. So do we.
Make you
2013/2/22 Peter Prettenhofer :
> http://xkcd.com/394/
Also http://xkcd.com/1000/
--
Lars Buitinck
Scientific programmer, ILPS
University of Amsterdam
--
Everyone hates slow websites. So do we.
Make your web apps faster
On 02/22/2013 12:03 PM, Christian wrote:
> Hi,
>
> when I train a classification model with feature selected data, I'll
> need for future scoring issues the selector object and the model object.
> So I'll must persist both ( i.e. with pickle ), right ?
Yes.
But the selector is just a mask of siz
Hi,
when I train a classification model with feature selected data, I'll
need for future scoring issues the selector object and the model object.
So I'll must persist both ( i.e. with pickle ), right ?
Many thanks
Christian
-
On 02/22/2013 11:39 AM, Andreas Mueller wrote:
> I was just wondering: does the current l1 penalty implementation actually
> lead to sparse coef_?
> I though additional tricks were required for that.
> If it is the case, maybe an example would be nice?
>
>
Oh, ok, the implementation indeed yields s
http://xkcd.com/394/
2013/2/22 Olivier Grisel :
> 2013/2/22 Peter Prettenhofer :
>> @ark: for 500K features and 3K classes your coef_ matrix will be:
>> 50 * 3000 * 8 / 1024. / 1024. ~= 11GB
>
> Nitpicking, that will be:
>
> 50 * 3000 * 8 / 1024. / 1024. ~= 11GiB
>
> or:
>
> 50 * 3000
2013/2/22 Peter Prettenhofer :
> @ark: for 500K features and 3K classes your coef_ matrix will be:
> 50 * 3000 * 8 / 1024. / 1024. ~= 11GB
Nitpicking, that will be:
50 * 3000 * 8 / 1024. / 1024. ~= 11GiB
or:
50 * 3000 * 8 / 1e6. ~= 12GB
But nearly everybody is making the mistake...
I was just wondering: does the current l1 penalty implementation actually
lead to sparse coef_?
I though additional tricks were required for that.
If it is the case, maybe an example would be nice?
On 02/22/2013 11:15 AM, Peter Prettenhofer wrote:
> I just opened a PR for this issue:
> https://gi
Hi Fredrik,
Given that OPTICS is a fairly standard clustering algorithm that can be
made efficient on large datasets, I do believe that it would be
interesting to have an implementation. Of course, the usual caveat apply:
we need high-quality, efficient, tested and well-documented code. It will
ta
I just opened a PR for this issue:
https://github.com/scikit-learn/scikit-learn/pull/1702
2013/2/22 Peter Prettenhofer :
> @ark: for 500K features and 3K classes your coef_ matrix will be:
> 50 * 3000 * 8 / 1024. / 1024. ~= 11GB
>
> Coef_ is stored as a dense matrix - you might get a considera
Is there any interest in an implementation of OPTICS [1] for
sklearn.cluster?
As part of our thesis work we've extended the cluster package to include
the OPTICS algorithm which returns an ordering and reachability distances
for the input samples. We're also planning on extracting actual clusters
@ark: for 500K features and 3K classes your coef_ matrix will be:
50 * 3000 * 8 / 1024. / 1024. ~= 11GB
Coef_ is stored as a dense matrix - you might get a considerable
smaller matrix if you use sparse regularization (keeps most
coefficients zero) and convert the coef_ array to a scipy sparse
17 matches
Mail list logo
