2013/8/9 Mathieu Blondel :
> This is not intentional. The question is why the reuters corpus was in the
> source directory ? It should have been in ~/scikit_learn_data in the first
> place. My guess is that the problem is in
> https://github.com/scikit-learn/scikit-learn/blob/master/examples/applic
Hi,
This is not intentional. The question is why the reuters corpus was in the
source directory ? It should have been in ~/scikit_learn_data in the first
place. My guess is that the problem is in
https://github.com/scikit-learn/scikit-learn/blob/master/examples/applications/plot_out_of_core_classi
Hi. Maybe this is a stupid observation, but release 0.14 is about 10 times the
size of release 0.13. Looking through the directories, I see
./examples/applications/reuters contains the Reuters-21578 corpus (35 megs),
both compressed and expanded. This looks like a mistake. Is this intentiona
Hi,
thanks to all who answered. using nosetests fixed all the tests error.
Congratulation for the release.
Frédéric
On Thu, Aug 8, 2013 at 11:53 AM, Olivier Grisel wrote:
> 2013/8/8 Gael Varoquaux :
> > On Thu, Aug 08, 2013 at 11:36:51AM -0400, Frédéric Bastien wrote:
> >> The way I run the t
Well done to everyone who worked on this.
I'll download and install tomorrow morning.
Regards,
Nigel Legg
07914 740972
http://www.trevanianlegg.co.uk
http://twitter.com/nigellegg
http://uk.linkedin.com/in/nigellegg
On 8 August 2013 00:18, Gael Varoquaux wrote:
> Hi Scikiters,
>
> I have tagge
2013/8/8 Gael Varoquaux :
> On Thu, Aug 08, 2013 at 11:36:51AM -0400, Frédéric Bastien wrote:
>> The way I run the tests is to start python, import sklearn, then execute
>> sklearn.test().
>
> That not the right way to run the tests, and it hasn't been for a while.
> You should do 'nosetests -x skl
2013/8/8 Nelle Varoquaux :
> Hi Frédéric,
>
> This is not a good way to run the tests. Please, use nose to run the tests,
> or type "make test" in the root folder of scikit-learn.
This method only works to test the project buily from the source
folder. It does not work if you install the officiall
Hi Frédéric,
This is not a good way to run the tests. Please, use nose to run the tests,
or type "make test" in the root folder of scikit-learn.
Thanks,
N
On 8 August 2013 17:36, Frédéric Bastien wrote:
> Hi,
>
> I just installed it and I have many tests errors. I also had the same sort
> of
On Thu, Aug 08, 2013 at 11:36:51AM -0400, Frédéric Bastien wrote:
> The way I run the tests is to start python, import sklearn, then execute
> sklearn.test().
That not the right way to run the tests, and it hasn't been for a while.
You should do 'nosetests -x sklearn' from the command line. In add
Hi,
I just installed it and I have many tests errors. I also had the same sort
of error with the 0.13 version. Here is a few example of the common error I
got:
==
ERROR: Doctest: sklearn.tree.tree.DecisionTreeClassifier
-
Hey everybody.
Thanks a lot and congratulations to everybody who worked on this :)
Cheers,
Andy
On 08/08/2013 01:18 AM, Gael Varoquaux wrote:
> Hi Scikiters,
>
> I have tagged and pushed release 0.14.
>
> Vlad will build the Windows binaries tomorrow. In the mean time, the 0.14
> tarball is on
I gather there's a lot of historical default behaviour in those metrics.
This results in a few quirks some of which are real problems, as in
https://github.com/scikit-learn/scikit-learn/issues/2094. It should be
cleaned up, but existing behaviour will need to be maintained for at least
two releases
Btw, micro-average is trivial to implement (assuming I understand this
correctly; I've not looked at the paper):
def multilabel_pr_curve(y_true, y_scores):
y_true = LabelBinarizer.fit_transform(y_true)
return precision_recall_curve(y_true.ravel(), y_score.ravel())
Macro-average is tricki
Congratulations !
Bertrand
- Mail original -
> De: "Gael Varoquaux"
> À: scikit-learn-general@lists.sourceforge.net
> Envoyé: Jeudi 8 Août 2013 01:18:09
> Objet: [Scikit-learn-general] Release 0.14: tagged and pushed!
>
> Hi Scikiters,
>
> I have tagged and pushed release 0.14.
>
> Vl
Thanks Gael for leading the final push on this release. The changelog is
indeed very impressive.
We can now focus on new stuff for 0.15 :)
--
Get 100% visibility into Java/.NET code with AppDynamics Lite!
It's a free troub
15 matches
Mail list logo
