Shubham,
There are many open improvements on the GitHub issues list
(https://github.com/scikit-learn/scikit-learn/issues?q=is%3Aopen+is%3Aissue+label%3AEasy).
I recommend starting with a few of the Easy or Documentation tasks -
it helps get the workflow down and is also very valuable to the
projec
Hi, I would like to start contributing to scikit learn. What are some tasks
a machine learning starter can begin with ? I am familiar with fundamental
machine learning algorithms like Naive Bayes, SVMs etc.
Thanks and regards,
Shubham
---
Scaling (or the same scaling procedure) is not always beneficial, but you
can certainly do exactly what you are saying by making a pipeline of a
StandardScaler and your estimator.
See the documentation for Pipeline at
http://scikit-learn.org/dev/modules/pipeline.html and
http://scikit-learn.org/de
So basically you would not do grid search, and just look at the proportion of
outliers in training and test datasets.
But what if you have so few examples of anomalies, that you want to keep them
all in the test dataset. In that case, how would you go about finding the best
value of nu?
Thank
Typically one should scale training data and then test data using the values
gotten when scaling training data.
When performing grid search, shouldn't scaling occur every time (k-1) folds are
selected as training data, and apply that to the k-th fold? So, in practice,
everytime the training/tes
2014-09-06 10:12 GMT+02:00 Gael Varoquaux :
> On Fri, Sep 05, 2014 at 10:12:14PM -0400, Sebastian Raschka wrote:
>> just saw that scikit-learn 15.2.0 is out and the LDA was fixed. That's great
>> :).
>
> I missed that reviewing the patches that made it into the 0.15.2.
>
> The change in LDA should
Alright I could not see the left panel with the version info on my
mobile phone version when I answered that email.
I now understand what happened: I buily the doc from the correct
source checkout (tag 0.15.2) but using the wrong version of sklearn in
my Python site-packages (I had master there be
Kyle Kelley reported that the source tarball that I uploaded yesterday
night on PyPI had a typo in its setup.py causing an import error. This
was caused by a mistake on my side when doing the release.
I regenerated the correct tarball from the 0.15.2 git tag and
reuploaded it to PyPI.
Sorry for a
On Sat, Sep 06, 2014 at 04:13:32PM +0200, Olivier Grisel wrote:
> > Looking at this page, however, there is a problem with the version
> > number: the page says that the docs are for 0.16, while they are for
> > 0.15.2.
> Where in the page on which URL ?
http://scikit-learn.org/stable/auto_exampl
Le 6 sept. 2014 10:12, "Gael Varoquaux" a
écrit :
>
> On Fri, Sep 05, 2014 at 10:12:14PM -0400, Sebastian Raschka wrote:
> > just saw that scikit-learn 15.2.0 is out and the LDA was fixed. That's
great :).
>
> I missed that reviewing the patches that made it into the 0.15.2.
>
> The change in LDA
Weird. I don't have access to my computer right now. Will need to check
what happened with the website when home. I thought I had synced the 0.15 /
stable root from the 0.15.2 tag.
--
Slashdot TV.
Video for Nerds. Stuff
Hi Roberto,
One possible way to tune the hyperparameters of the One Class SVM is to
split the data set in training and test sets, train the One Class SVM with
the training set and a pre-specified nu, and see if you get a
similar amount of proportion of outliers (a number close to nu) on the test
s
On Fri, Sep 05, 2014 at 10:12:14PM -0400, Sebastian Raschka wrote:
> just saw that scikit-learn 15.2.0 is out and the LDA was fixed. That's great
> :).
I missed that reviewing the patches that made it into the 0.15.2.
The change in LDA should _not_ have made to to a minor bugfix release. A
chang
On Sat, Sep 06, 2014 at 02:26:31AM +0200, Olivier Grisel wrote:
> I just released 0.15.2. The source and binary packages for this
> release are on PyPi as usual:
> https://pypi.python.org/pypi/scikit-learn/0.15.2
Congratulations! And thanks you in the name of our users. This provides
plenty of va
14 matches
Mail list logo
