Timmy,
Interesting... this means that none of the earlier ideas were on the
right track, or else we'd see the same error for standard and modified.
It turns out that Hessian and LTSA do very similar things. I still
haven't wrapped my mind around it intuitively, but what you're seeing
probably
'standard' + 'modified' both work fine
'hessian' + 'ltsa' both have issues
ltsa is printing:
RuntimeWarning: Diagonal number 2 is exactly zero. Singular matrix.
and then setting everything to -nan
i tried adding random noise and increasing n_neighbors -- but no dice
Fabian's suggestion is sti
On Thu, Dec 8, 2011 at 2:35 AM, Olivier Grisel wrote:
> Ok I have edited the wiki page to add scope, motivation and use case
> information for this sprint.
>
> Please feel free to edit and comment:
>
> http://wiki.ipython.org/PyCon12Sprint
>
> In particular I am interested in feedback on which us
I'm not a contributor, but +1 for the drop of 2.5 compatibility
(reasoning below)
On Thu, Dec 8, 2011 at 7:34 AM, Yaroslav Halchenko wrote:
> it might be handy to know which particular features of 2.6 are you
> aiming to use which aren't in 2.5... ?
one important point for dropping 2.5: it's far
Thanks Vlad, I did just book a room. Very cheap and looks nice. Looking
forward to meeting you guys!
David
On Thu, Dec 08, 2011 at 10:33:54PM +0200, Vlad Niculae wrote:
>
> On Dec 8, 2011, at 20:11 , David Warde-Farley wrote:
>
> > On Tue, Nov 15, 2011 at 03:13:53AM +0900, Mathieu Blondel wrote
On 9 December 2011 04:52, Olivier Grisel wrote:
> Also Python 2.5 is no longer supported by the PSF, (since May 2011)
> even for security fixes:
>
> Python 2.5.6
>
>We are pleased to announce the release of Python 2.5.6, a security
> fix release of Python 2.5, on May 26th, 2011.
>
> This is a
On Dec 8, 2011, at 20:11 , David Warde-Farley wrote:
> On Tue, Nov 15, 2011 at 03:13:53AM +0900, Mathieu Blondel wrote:
>> Hi,
>>
>> Thanks heaps Gael. I'm planning to contact the guy by tomorrow. I
>> think it would be easier for him if we don't contact him individually.
>> I can make the reser
On Tue, Nov 15, 2011 at 03:13:53AM +0900, Mathieu Blondel wrote:
> Hi,
>
> Thanks heaps Gael. I'm planning to contact the guy by tomorrow. I
> think it would be easier for him if we don't contact him individually.
> I can make the reservations for the remaining people who don't have a
> reservatio
Also Python 2.5 is no longer supported by the PSF, (since May 2011)
even for security fixes:
Python 2.5.6
We are pleased to announce the release of Python 2.5.6, a security
fix release of Python 2.5, on May 26th, 2011.
This is a source-only release that only includes security fixes. The
last
rright -- thanks Josef,
just for completeness
$> cat /tmp/1.py
def f(arg1, arg2=1, **kwargs):
print "got ", arg1, arg2, kwargs
f(0, *(10,), buga=1)
$> python2.5 /tmp/1.py
File "/tmp/1.py", line 4
f(0, *(10,), buga=1)
^
SyntaxError: invalid syntax
$> python2.6 /tm
I reduced the problem to a difference of results with eigsh when
working in shift-invert mode and when not. Notice that the following
works with sigma=None but not with sigma=0. , thus seems to me that we
should fall back to non invert-mode for singular matrices (see
asociated pull request).
```
I
2011/12/8 Gael Varoquaux :
> Yes. We'll probably be able to clean our backports in sklearn/utils. I
> look forward to removing code.
https://github.com/scikit-learn/scikit-learn/pull/458
But there seems to be more scipy 0.7.0 compat stuff left for you :)
--
Lars Buitinck
Scientific programmer,
On Thu, Dec 08, 2011 at 11:21:56AM -0500, David Warde-Farley wrote:
> Indeed, while this might not be there, there is a lot of useful new stuff in
> collections and itertools in 2.6 that may come in handy.
Yes. We'll probably be able to clean our backports in sklearn/utils. I
look forward to rem
Sorry for the empty reply, clumsy phone fingers...
On 2011-12-08, at 11:00, Lars Buitinck wrote:
> 2011/12/8 Gael Varoquaux :
>> On Thu, Dec 08, 2011 at 10:34:44AM -0500, Yaroslav Halchenko wrote:
>>> it might be handy to know which particular features of 2.6 are you
>>> aiming to use which aren
Sent from my iPhone
On 2011-12-08, at 11:00, Lars Buitinck wrote:
> 2011/12/8 Gael Varoquaux :
>> On Thu, Dec 08, 2011 at 10:34:44AM -0500, Yaroslav Halchenko wrote:
>>> it might be handy to know which particular features of 2.6 are you
>>> aiming to use which aren't in 2.5... ?
>>
>> Named t
On 2011-12-08, at 10:12, Gael Varoquaux wrote:
> On Thu, Dec 08, 2011 at 04:09:00PM +0100, Olivier Grisel wrote:
>> Gael suggested on a checkins comment on github to drop support for
>> Python 2.5. Let's vote. Besides what is you opinion on oldest numpy /
>> scipy supported version.
>
> I'd like
2011/12/8 Lars Buitinck :
> collections.Counter
This was introduced in 2.7, actually. But the rest of my wish list is
there in 2.6 :)
--
Lars Buitinck
Scientific programmer, ILPS
University of Amsterdam
--
Cloud Service
On Thu, Dec 8, 2011 at 10:51 AM, Gael Varoquaux
wrote:
> On Thu, Dec 08, 2011 at 10:44:23AM -0500, Yaroslav Halchenko wrote:
>> > and function calls of type 'f(arg1, arg2=1, **kwargs)'.
>> hm... not sure what you mean (there was some change on how keyword args
>> are handled but can't recall now),
On Thu, Dec 08, 2011 at 05:00:57PM +0100, Lars Buitinck wrote:
> class decorators
Yes, I've missed class decorators so much.
> If we want, we can even do from __future__ import print_function
I could go with that.
G
-
2011/12/8 Gael Varoquaux :
> On Thu, Dec 08, 2011 at 10:34:44AM -0500, Yaroslav Halchenko wrote:
>> it might be handy to know which particular features of 2.6 are you
>> aiming to use which aren't in 2.5... ?
>
> Named tuples, and function calls of type 'f(arg1, arg2=1, **kwargs)'.
Stuff I've miss
On Thu, Dec 8, 2011 at 10:34 AM, Yaroslav Halchenko wrote:
> it might be handy to know which particular features of 2.6 are you
> aiming to use which aren't in 2.5... ?
>
I'm not a contributor to scikit-learn, but I would use the
getter/setter/deleter decorators for properties, if I didn't have t
On Thu, Dec 08, 2011 at 10:44:23AM -0500, Yaroslav Halchenko wrote:
> > and function calls of type 'f(arg1, arg2=1, **kwargs)'.
> hm... not sure what you mean (there was some change on how keyword args
> are handled but can't recall now), since following snippet works in both
OK, so I got it wrong
On Thu, 08 Dec 2011, Gael Varoquaux wrote:
> > aiming to use which aren't in 2.5... ?
> Named tuples,
yeah
> and function calls of type 'f(arg1, arg2=1, **kwargs)'.
hm... not sure what you mean (there was some change on how keyword args
are handled but can't recall now), since following snippet
On 12/08/2011 04:34 PM, Yaroslav Halchenko wrote:
> it might be handy to know which particular features of 2.6 are you
> aiming to use which aren't in 2.5... ?
>
As I am not so familiar with the Python versions, I
was also wondering that.
I'd like to pitch the versions in latest Ubuntu LTS: 10.04:
On Thu, Dec 08, 2011 at 10:34:44AM -0500, Yaroslav Halchenko wrote:
> it might be handy to know which particular features of 2.6 are you
> aiming to use which aren't in 2.5... ?
Named tuples, and function calls of type 'f(arg1, arg2=1, **kwargs)'.
In general, it is also about lower the bunden of
it might be handy to know which particular features of 2.6 are you
aiming to use which aren't in 2.5... ?
On Thu, 08 Dec 2011, Olivier Grisel wrote:
> Gael suggested on a checkins comment on github to drop support for
> Python 2.5. Let's vote. Besides what is you opinion on oldest numpy /
> scipy
On Thu, Dec 08, 2011 at 04:09:00PM +0100, Olivier Grisel wrote:
> Gael suggested on a checkins comment on github to drop support for
> Python 2.5. Let's vote. Besides what is you opinion on oldest numpy /
> scipy supported version.
I'd like to pitch the versions in latest Ubuntu LTS: 10.04:
Pytho
>
> [X] +1
>
--
Cloud Services Checklist: Pricing and Packaging Optimization
This white paper is intended to serve as a reference, checklist and point of
discussion for anyone considering optimizing the pricing and packagi
2011/12/8 Olivier Grisel :
>
> Vote for dropping Python 2.5 support: supported versions will be 2.6.x
> and 2.7.x + 3.2+ (at some point).
> [X] +1
--
Olivier
http://twitter.com/ogrisel - http://github.com/ogrisel
--
Clou
Hi list,
Gael suggested on a checkins comment on github to drop support for
Python 2.5. Let's vote. Besides what is you opinion on oldest numpy /
scipy supported version.
Currently Fabian tests the releases on python 2.5, numpy 1.3.0 and
scipy 0.7.2. Once we decide on the lowest supported version
On Thu, Dec 08, 2011 at 07:02:01AM -0800, Jacob VanderPlas wrote:
> Sometimes this sort of error can happen when not enough neighbors are
> used, so that your data is split into two unconnected regions.
> Increasing the number of neighbors could help
Maybe we could give this advice in the trace
2011/12/8 María Helena Mejía Salazar :
>
>
> --
> Cloud Services Checklist: Pricing and Packaging Optimization
> This white paper is intended to serve as a reference, checklist and point of
> discussion for anyone consideri
Sometimes this sort of error can happen when not enough neighbors are
used, so that your data is split into two unconnected regions.
Increasing the number of neighbors could help
Jake
Timmy Wilson wrote:
> I get the following error when running Hessian-based LLE::
>
>
> /home/timmyt/projects/
--
Cloud Services Checklist: Pricing and Packaging Optimization
This white paper is intended to serve as a reference, checklist and point of
discussion for anyone considering optimizing the pricing and packaging model
of
Hi everyone,
I've edited the wiki with ideas for the two main tasks I'm planning to
work on during the sprint (K-means improvements and Random
Projections).
https://github.com/scikit-learn/scikit-learn/wiki/Upcoming-events
Feel free to add the tasks you plan to work on, so that we can avoid
over
On Thu, Dec 8, 2011 at 2:48 AM, Timmy Wilson wrote:
> I get the following error when running Hessian-based LLE::
>
>
Hi Timmy,
Please check if this solves your problem:
https://github.com/scikit-learn/scikit-learn/pull/457
Best,
Fabian
2011/12/8 Satrajit Ghosh :
> hi olivier,
>
> thanks for the info. that is very helpful. i would be very curious how you
> plan to make efficient data distribution across nodes of a cluster and
> balance i/o performance. (or is part of the low overhead the notion that you
> are looking at a shared m
hi olivier,
thanks for the info. that is very helpful. i would be very curious how you
plan to make efficient data distribution across nodes of a cluster and
balance i/o performance. (or is part of the low overhead the notion that
you are looking at a shared memory architecture primarily).
cheers
Ok I have edited the wiki page to add scope, motivation and use case
information for this sprint.
Please feel free to edit and comment:
http://wiki.ipython.org/PyCon12Sprint
In particular I am interested in feedback on which use cases are the
most important to you.
--
Olivier
--
39 matches
Mail list logo
