Hi everyone,
I was wondering what the LogisticRegression class in
sklearn.linear_modules implements. Its docstring says just logistic
regression, but I stumbled upon the liblinear authors' paper on LR
[1], and they claim to have developed a fast multiclass LR/MaxEnt
training algorithm as well.
Is there a function in sklearn that will calculate non-regularized logistic
regression? or should I just pick a really large C parameter for
linear_model.LogisticRegression?
Thanks in advance.--
All the data
LR and ME have a strong relation. I remember there was once a discussion in
this list. But I am not sure what is actually implemented in scikit.learn.
On Wed, Oct 19, 2011 at 5:32 PM, Lars Buitinck l.j.buiti...@uva.nl wrote:
Hi everyone,
I was wondering what the LogisticRegression class in
On Wed, Oct 19, 2011 at 05:32, Lars Buitinck l.j.buiti...@uva.nl wrote:
Hi everyone,
I was wondering what the LogisticRegression class in
sklearn.linear_modules implements. Its docstring says just logistic
regression, but I stumbled upon the liblinear authors' paper on LR
[1], and they claim
On Wed, Oct 19, 2011 at 6:32 PM, Lars Buitinck l.j.buiti...@uva.nl wrote:
I was wondering what the LogisticRegression class in
sklearn.linear_modules implements. Its docstring says just logistic
regression, but I stumbled upon the liblinear authors' paper on LR
[1], and they claim to have
On Wed, Oct 19, 2011 at 8:37 PM, Alexandre Passos
alexandre...@gmail.com wrote:
Maxent is usually older terminology that used to be favored in the NLP
community but has recently been phased out due to confusion and a
mismatch with the way people actually implement things.
Many NLP people use
On Wed, Oct 19, 2011 at 8:36 PM, Gael Varoquaux
gael.varoqu...@normalesup.org wrote:
AFAIK the only way it is to pick a large C. In my experience this works
well.
Or alpha=0 in SGD.
Mathieu
--
All the data
I agree, a lot of people in NLP use the term MaxEnt to denote
multinomial logistic regression.
2011/10/19 Mathieu Blondel math...@mblondel.org:
On Wed, Oct 19, 2011 at 8:37 PM, Alexandre Passos
alexandre...@gmail.com wrote:
Maxent is usually older terminology that used to be favored in the
On Tue, Oct 18, 2011 at 11:35 PM, xinfan meng mxf3...@gmail.com wrote:
Hi. This is not directly related to the codes but I found this
confusing. The paper proposed voted perceptron is called Large Margin
Classification
Using the Perceptron Algorithm and the docs fo voted perceptron also
I think we should put a long summary of this discussion in the
narrative doc and a short summary in the docstring of the class.
--
All the data continuously generated in your IT infrastructure contains a
definitive record
2011/10/19 Olivier Grisel olivier.gri...@ensta.org:
I think we should put a long summary of this discussion in the
narrative doc and a short summary in the docstring of the class.
Yes, by MaxEnt I meant the true multiclass version of LR rather than
the OvA version. I put a note to this effect
On Tue, Oct 18, 2011 at 7:49 PM, Alexandre Gramfort
alexandre.gramf...@inria.fr wrote:
could you open an issue with a small test script with one X and y that
produce a different result using both implementations?
Here it is:
https://github.com/scikit-learn/scikit-learn/issues/403
I notice
On Wed, Oct 19, 2011 at 8:42 AM, Lars Buitinck l.j.buiti...@uva.nl wrote:
Yes, by MaxEnt I meant the true multiclass version of LR rather than
the OvA version. I put a note to this effect in the docstring, but
I'll leave the narrative docs to someone who is more knowledgeable
than me wrt.
This is cross-posted from the scikits.image mailing list; It was so
interesting, I thought it a waste not to use the opportunity.
We've had a number of discussions on cython types, and how we wish that
cython would support some sort of templates. This would be very useful for
the `tree` module
Interesting. I've been staring at the code but the algorithm itself
shouldn't be losing precision. On the other hand, there are those
stopping conditions that I had taken from the C implementation of
the author of the Cholesky-OMP paper. If it's as you say, it could be
that when it fails, OMP
Hey all,
I got tired of typing the long sourceforge address, so I've registered
http://scikit-learn.org
and had it do a 301 redirect to http://scikit-learn.sourceforge.net.
Of course, this domain really belongs to the project, so let me know
if you need a different configuration.
The
Hi all,
In the auto-docs found at
http://scikit-learn.sourceforge.net/stable/modules/classes.html,
anything using :template: function.rst for the auto-summary does not
have a link to the documentation. (e.g, almost everything under the
heading Datasets heading on the page). Items using
Hi Jake,
A while back I remember having that issue because my local version of
sphinx was higher than 1.0.0 and thus unsupported by the scikit-learn
docs, so the function links wouldn't work when I built it locally, but
they would work in the online-generated version.
Did the sphinx version used
2011/10/19 Stéfan van der Walt ste...@sun.ac.za:
Hey all,
I got tired of typing the long sourceforge address, so I've registered
http://scikit-learn.org
and had it do a 301 redirect to http://scikit-learn.sourceforge.net.
Of course, this domain really belongs to the project, so let me know
2011/10/19 Brian Holt bdho...@gmail.com:
This is cross-posted from the scikits.image mailing list; It was so
interesting, I thought it a waste not to use the opportunity.
We've had a number of discussions on cython types, and how we wish that
cython would support some sort of templates. This
The BallTree cython code would be a quick way to test this: the
variables DTYPE and DTYPE_t are defined at the top, and all that would
need to be templated are these definitions. Though come to think of it,
I haven't yet tested compilation with a different DTYPE...
Jake
Olivier Grisel
On Wed, Oct 19, 2011 at 06:25:03PM +0100, Brian Holt wrote:
We've had a number of discussions on cython types, and how we wish that
cython would support some sort of templates. This would be very useful
for the `tree` module (that is currently tied to a 32bit implementation)
and a
2011/10/19 Gael Varoquaux gael.varoqu...@normalesup.org:
On Wed, Oct 19, 2011 at 06:25:03PM +0100, Brian Holt wrote:
We've had a number of discussions on cython types, and how we wish that
cython would support some sort of templates. This would be very useful
for the `tree` module
On Wed, Oct 19, 2011 at 11:33:44AM -0700, Stéfan van der Walt wrote:
I got tired of typing the long sourceforge address, so I've registered
http://scikit-learn.org
Awesome.
We'll need to change our cname on our sourceforge page to show up as
scikit-learn.org. I don't know how to do that, and
On Wed, Oct 19, 2011 at 22:26, Gael Varoquaux
gael.varoqu...@normalesup.org wrote:
On Wed, Oct 19, 2011 at 06:25:03PM +0100, Brian Holt wrote:
We've had a number of discussions on cython types, and how we wish that
cython would support some sort of templates. This would be very useful
On Wed, Oct 19, 2011 at 10:43:52PM +0100, Robert Kern wrote:
I recommend Tempita, a tiny but well-featured templating engine that
can be dropped into your codebase:
http://pythonpaste.org/tempita/
Very nice, I was actually looking for something like that.
Robert, do you have a twitter
On Wed, Oct 19, 2011 at 22:48, Gael Varoquaux
gael.varoqu...@normalesup.org wrote:
On Wed, Oct 19, 2011 at 10:43:52PM +0100, Robert Kern wrote:
I recommend Tempita, a tiny but well-featured templating engine that
can be dropped into your codebase:
http://pythonpaste.org/tempita/
Very
On Wed, Oct 19, 2011 at 2:35 PM, Gael Varoquaux
gael.varoqu...@normalesup.org wrote:
We'll need to change our cname on our sourceforge page to show up as
scikit-learn.org. I don't know how to do that, and I bet that I don't
have the rights, but I am sure that Fabian knows.
Instructions from
2011/10/19 Stéfan van der Walt ste...@sun.ac.za:
Instructions from SF:
- Login to SourceForge.net as a project administrator and go to the
Project Summary page for your project.
- Go to the Project Admin pulldown menu.
- Click Feature Settings.
- Find the Virtual Hosts (VHOSTs) row.
-
On 2011-10-19, at 13:12, Jason Rennie jren...@gmail.com wrote:
On Wed, Oct 19, 2011 at 8:42 AM, Lars Buitinck l.j.buiti...@uva.nl wrote:
Yes, by MaxEnt I meant the true multiclass version of LR rather than
the OvA version. I put a note to this effect in the docstring, but
I'll leave the
30 matches
Mail list logo
