As there is so much positive feedback, I might make something up tonight.
As I like to make small steps, I'd get rid of the defunc search bar and
add some more
menu items instead (and adjust the respective pages obv.)
ps: For those who are wondering: no, I didn't choose this time because
Gael
For Windows, installing numba is a breeze using:
http://www.lfd.uci.edu/~gohlke/pythonlibs/
Basically, all the gnarly extensions are available already compiled with
all the dependencies handled properly. It's absolutely amazing and I
strongly encourage everyone who uses Python on Windows
+1
On Tue, Mar 5, 2013 at 4:19 PM, Alexandre Gramfort
alexandre.gramf...@inria.fr wrote:
ps: For those who are wondering: no, I didn't choose this time because
Gael is offline.
:)
I'd rather like to have his input :-/ he is back in a month, right?
3 weeks
Alex
So are you saying llvm isn't needed, if numba/llvmpy are installed from
Christoph's packages?
On Tue, Mar 5, 2013 at 9:21 AM, federico vaggi vaggi.feder...@gmail.comwrote:
For Windows, installing numba is a breeze using:
http://www.lfd.uci.edu/~gohlke/pythonlibs/
Basically, all the gnarly
On 03/05/2013 09:11 AM, Andreas Mueller wrote:
As there is so much positive feedback, I might make something up tonight.
As I like to make small steps, I'd get rid of the defunc search bar and
add some more
menu items instead (and adjust the respective pages obv.)
I made a page but the CSS
Yup - you can just install those packages, then try to run the default
example/tests, and both pass for me! For other packages, like mysqldb,
which is a breeze to compile on Linux, but compiling it on Windows under 64
bit is incredibly painful. Here is a good guide if you want to do it on
your
Ok, working now:
http://amueller.github.com/
All feedback welcome :)
I'd like to avoid bombarding the user with long lists / pages as much as
possible.
The Getting Started and Development pages now are a length that
mostly fit on
a screen and that I can still grasp.
If we had an algorithms page
This looks good. Maybe we could reintroduce a canonical snippet on the
home page:
from sklearn.datasets import load_digits
from sklearn.cross_validation import train_test_split
from sklearn.svm import LinearSVC
digits = load_digits()
X_train, X_test, y_train, y_test = train_test_split(
...
Exactly. Not only would you need cython, it also needs to be a recent version.
people with older versions would get cryptic error messages, leading to
frustrated users and busy mailing lists.
Matthieu Brucher matthieu.bruc...@gmail.com schrieb:
Hi,
If I remember correctly, this is done to
hi all. just looking at the one class svm and I'd like to get a
probabililty rather than a distance output. i know that in regular
svms you can get parameters for the sigmoid function from five-fold
cross validation and that's done by setting the probability=True in
the constructor. i presume it's
2013/3/5 Bill Power bill.power...@gmail.com:
investigating previous versions i saw that probability was available
in version 0.9 with predict_proba and predict_log_proba functions
http://scikit-learn.org/0.9/modules/generated/sklearn.svm.OneClassSVM.html
but it's not here in the stable
thanks lars
i figured as much. do you know if there are any ppaers in the
literature that i might be able to implement and then perhaps
contribute the code to the package? or do i have to live with either
using distances or a non-parameterised sigmoid function?
thanks
libsvm does not support probability outputs for one-class SVM.
One-class SVM is an algorithm for support estimation (not proper
density estimation) - i.e. you get a confidence that P(X) t - where
t is somewhat concealed in the nu parameter.
2013/3/5 Lars Buitinck l.j.buiti...@uva.nl:
2013/3/5
I like your changes Andy. It's definitely easier to navigate. I'm currently
also changing your graph from
http://peekaboo-vision.blogspot.de/2013/01/machine-learning-cheat-sheet-for-scikit.htmlinto
a documentation-linking version that can be added to the
documentation. I'll try put an online build
I feel like the About us section on the homepage shouldn't be there.
I'd rather put a About link somewhere else than putting this in
front on the home page. Also, I would use the space that we now have
on the front page to highlight more important aspects of the package.
On 5 March 2013 14:46,
On 03/05/2013 02:55 PM, Gilles Louppe wrote:
I feel like the About us section on the homepage shouldn't be there.
I'd rather put a About link somewhere else than putting this in
front on the home page. Also, I would use the space that we now have
on the front page to highlight more important
On 03/05/2013 02:46 PM, Jaques Grobler wrote:
I like your changes Andy. It's definitely easier to navigate. I'm
currently also changing your graph from
http://peekaboo-vision.blogspot.de/2013/01/machine-learning-cheat-sheet-for-scikit.html
into a documentation-linking version that can be
can you make the new website design part of a repo so we can submit PRs or
issues against it?
On Tue, Mar 5, 2013 at 9:39 AM, Andreas Mueller amuel...@ais.uni-bonn.dewrote:
On 03/05/2013 03:18 PM, Nelle Varoquaux wrote:
Hi everyone,
I'm actually not convinced about the new layout (sorry
On 03/05/2013 03:46 PM, Ronnie Ghose wrote:
can you make the new website design part of a repo so we can submit
PRs or issues against it?
It is a branch in my sklearn fork, but the branch is not completely up
to date, working on it.
Hi,
What clustering technique (with implementation in sklearn) is recommended
for 1d data?
--
Everyone hates slow websites. So do we.
Make your web apps faster with AppDynamics
Download AppDynamics Lite for free today:
..does kmeans not work?
On Tue, Mar 5, 2013 at 9:51 AM, nipun batra nipunredde...@gmail.com wrote:
Hi,
What clustering technique (with implementation in sklearn) is recommended
for 1d data?
--
Everyone
On 03/05/2013 03:51 PM, nipun batra wrote:
Hi,
What clustering technique (with implementation in sklearn) is
recommended for 1d data?
I'd recommend looking at it ;)
It feels like there might be some sweeping algorithm to get the optimal
solution for the k-means algorithm.
KMeans should be
It should. I would have straight away tried it, but read the following 2
posts:
1. http://stackoverflow.com/questions/11513484/1d-number-array-clustering
2. http://stats.stackexchange.com/questions/13781/clustering-1d-data
Any thoughts?
On Tue, Mar 5, 2013 at 8:24 PM, Ronnie Ghose
On 03/05/2013 03:18 PM, Nelle Varoquaux wrote:
Hi everyone,
I'm actually not convinced about the new layout (sorry Andy :( ). I
should also say, I'm not convinced about panda's website.
The menu is, I think, quite confusing. Overall, I think there are two
many links which may refer to the
On 5 March 2013 15:39, Andreas Mueller amuel...@ais.uni-bonn.de wrote:
On 03/05/2013 03:18 PM, Nelle Varoquaux wrote:
Hi everyone,
I'm actually not convinced about the new layout (sorry Andy :( ). I
should also say, I'm not convinced about panda's website.
The menu is, I think, quite
Awesome, thanks. You have my inkscape file, right?
I am trying to make the user guide also more flat by putting your java
script function in the file now :)
I think as the user guide is much shorter now, it doesn't really hide
things, but rather makes them easier to find.
We'll see.
@Andy
Congratulations :) Nice work
2013/3/4 Johannes Schönberger jschoenber...@demuc.de
Announcement: scikit-image 0.8.0
We're happy to announce the 8th version of scikit-image!
scikit-image is an image processing toolbox for SciPy that includes
algorithms
for
interesting posts :).
so
1) do we want a natural breaks method?
https://en.wikipedia.org/wiki/Jenks_natural_breaks_optimization
2) have you considered looking at the distribution of the variable as they
suggest? any small-d tends to allow this rather than the usual giant-d
space.
Do you have any
On 03/05/2013 04:04 PM, Nelle Varoquaux wrote:
Maybe that is the problem the core problem. The documentation has not
been written to be without sections: before, the user guide was
divided into three parts:
1. Installation
2. Tutorials: an overview of the scikit
3. Unsupervised learning
4.
On 03/05/2013 04:11 PM, Jaques Grobler wrote:
Awesome, thanks. You have my inkscape file, right?
I am trying to make the user guide also more flat by putting your java
script function in the file now :)
I think as the user guide is much shorter now, it doesn't really hide
things, but
Hi Tom,
recently I saw the arff-package in pypi. Seems working.
import arff
import numpy as np
barray = []
for row in arff.load('/home/chris/tools/weka-3-7-6/rd54_train.arff'):
barray.append(list(row))
nparray = np.array(barray)
print nparray.shape
(4940, 56)
HTH
Christian
I’m trying
On 03/05/2013 04:04 PM, Nelle Varoquaux wrote:
Maybe that is the problem the core problem. The documentation has not
been written to be without sections: before, the user guide was
divided into three parts:
1. Installation
2. Tutorials: an overview of the scikit
3. Unsupervised learning
On 03/05/2013 04:11 PM, Jaques Grobler wrote:
Awesome, thanks. You have my inkscape file, right?
I am trying to make the user guide also more flat by putting your java
script function in the file now :)
I think as the user guide is much shorter now, it doesn't really hide
things, but
We really should have this support within the library. Does it make sense
to just use the functionality in the arff-package?
On Tue, Mar 5, 2013 at 7:34 AM, Christian mining.fa...@gmail.com wrote:
Hi Tom,
recently I saw the arff-package in pypi. Seems working.
import arff
import numpy as
2013/3/5 Rob Zinkov cfour...@gmail.com:
We really should have this support within the library. Does it make sense to
just use the functionality in the arff-package?
Is it better than the one in scipy.io?
--
Lars Buitinck
Scientific programmer, ILPS
University of Amsterdam
The fastcluster project by Dan Mullner, a professor of math and statistics at
Stanford, might be of interest.
http://math.stanford.edu/~muellner/fastcluster.html
These routines follow the same API of the hierarchical clustering routines in
scipy, including single linkage and complete linkage,
2013/3/5 Robert McGibbon rmcgi...@gmail.com:
The fastcluster project by Dan Mullner, a professor of math and statistics
at Stanford, might be of interest.
http://math.stanford.edu/~muellner/fastcluster.html
These routines follow the same API of the hierarchical clustering routines
in scipy,
Thanks for your response, Christian. I experimented with the package. FYI,
there’s a problem with the pypi arff reader. The package claims to handle
numbers but it seems to encode everything (including numbers) as strings, like
this:
[['blonde' '17.2' '1' 'yes']
['blue' '27.2' '2' 'yes']
On Mar 5, 2013, at 10:10 AM, Olivier Grisel wrote:
This code is in C++ and the scikit-learn core maintainers are not all
experts in C++ and prefer cython for optimized code.
A cython rewrite of some of those algorithms would be of interest though.
For anyone interested in either
The import method doesn't support sparse representations:
This function should be able to read most arff files. Not implemented
functionality include:
date type attributes
string type attributes
It can read files with numeric and nominal attributes. It cannot read files
with sparse data ({}
Hello everyone,
I am new to scikit-learn package, I am still trying sone of the examples on the
website.
You have an example on RBF SVM parameters that is very interesting but my only
problem is that my data are in libsvm format
I know that you have an option for loading this format through
Greetings,
I have read the developer guidelines for scikit learn, and I would like
to contribute (to boost my machine learning and python fu). Is there an
outstanding, easy bug or feature that can be assigned to me or should
I select one?
Thanks,
Jeff Van Voorst
http://stackoverflow.com/questions/13590247/using-libsvm-format-in-scikit
SO is amazing :)
On Tue, Mar 5, 2013 at 2:19 PM, Mohamed Radhouane Aniba
arad...@gmail.comwrote:
Hello everyone,
I am new to scikit-learn package, I am still trying sone of the examples
on the website.
You have an
Hi Jeff.
Thanks for your will to contribute.
As the dev guidelines state, there are certain issues that are tagged as
easy:
https://github.com/scikit-learn/scikit-learn/issues?labels=Easypage=1sort=updatedstate=open
These might still vary a lot. Maybe just browse around.
How familiar are you with
So I've finally got something to show. Gael, you were entirely correct
about it being a mouthful. I've been developing it as a separate package
for simplicity, but will be integrating with scikit-learn as soon as I get
the time. Here is what I've got so far in case anyone wants to take a look:
It was a pretty easy build on Mac -- I just used MacPorts to install and
select an llvm. Of course Anaconda is even easier.
I'd say Numba is a medium-term consideration. It's enough trouble getting
everybody using C compilers, so adding LLVM to the mix is probably way too
much of a change for the
On 03/05/2013 08:15 PM, Jason Rudy wrote:
So I've finally got something to show. Gael, you were entirely
correct about it being a mouthful. I've been developing it as a
separate package for simplicity, but will be integrating with
scikit-learn as soon as I get the time. Here is what I've
On 6 March 2013 06:26, Andreas Mueller amuel...@ais.uni-bonn.de wrote:
Hi Jeff.
Thanks for your will to contribute.
As the dev guidelines state, there are certain issues that are tagged as
easy:
https://github.com/scikit-learn/scikit-learn/issues?labels=Easypage=1sort=updatedstate=open
For me it works fine.
Cheers, Christian
test.arff
@relation 'test'
@attribute v1 {blonde,blue}
@attribute v2 numeric
@attribute v3 numeric
@attribute class {yes,no}
@data
blonde,17.2 ,1,yes
blue,27.2,2,yes
blue,18.2,3,no
end test.arff
barray
[['blonde', 17.2, 1.0, 'yes'],
['blue', 27.2,
Hi, I added a comment to issue #1158 but since it is closed, I'm not sure
if anyone would be alerted.
I am not sure if this should be closed or perhaps a second issue should be
opened.
As already stated, the attribute n_symbols only gets created when an
emission probability matrix is defined.
Hi.
Should we just deprecate / remove the HMM?
We deemed sequence prediction off-topic (Lars' words and I agree) and
there is no core-dev maintaining them.
Is there any project this could move to?
Statsmodel, pandas? There should be a go-to place for time-series modelling.
There was
Hi everybody.
The dns seems to be down again.
Should we try to switch? Does Stefan still have the domain or did
someone else take care of it?
Cheers,
Andy
--
Symantec Endpoint Protection 12 positioned as A LEADER in The
52 matches
Mail list logo