Hi,
Nice!
As for your previous message, I'd like to start working on a PR with
interpolation on the residuals. I don't have that much time on my
hands right now, but I keep this PR in the back of my head (I've seen
your email your holidays).
I think that even with this improved docstring it is sti
On Fri, Feb 15, 2013 at 12:53 PM, Olivier Grisel
wrote:
> How do you evaluate which value of alpha is the best if you don't
> cross validate in one way or another?
Just choose the alpha (from a fixed set) that minimizes the RMSE of
the prediction of the last time step (or the last n time steps wi
On Fri, Feb 15, 2013 at 11:30 AM, Gael Varoquaux
wrote:
> On Fri, Feb 15, 2013 at 11:22:02AM +0100, Charles-Pierre Astolfi wrote:
>> > However, before we do this, I'd like to understand: what is the usecase
>> > and the purpose for this function?
>> Whic
On Fri, Feb 15, 2013 at 10:35 AM, Gael Varoquaux
wrote:
>> WDYT?
> I find that lars_path is already a function that is complicated-enough,
> with a signature difficult to understand, so I'd rather not make it more
> complex. If there is a need, we can add a similar function that calls
> lars_path
inearModel (or LassoLars) would make more sense.
--
Cp
On Fri, Feb 15, 2013 at 12:29 AM, Olivier Grisel
wrote:
> 2013/2/14 Charles-Pierre Astolfi :
>> Hi everyone,
>>
>> A few months ago, I posted on this list about how
>> lars_path(method='lasso') was
Hi everyone,
A few months ago, I posted on this list about how
lars_path(method='lasso') was much faster than lasso_path, when one
has to compute lasso for many different regularization parameters
(alphas).
Basically the idea is that lars_path(method='lasso') returns all the
kinks (hitting points
I'm a noob when it comes to data on a sphere, but is there any issue
with preprocessing the data to project it on a place, run kmeans in
the plane and the reproject it back on the sphere?
There's no projection that conserves the distance wrt to any pair of
points on the sphere (although there are
Hi Gael!
I used to work (6 months internship) with Gilles Stoltz from the
CLASSIC team, but I was the only one using scikit-learn.
Cheers
--
Cp
On Thu, Jan 17, 2013 at 1:26 PM, Gael Varoquaux
wrote:
> If you are in an INRIA team that is using scikit learn, please tell me
> now, I need to list
alpha=0
>
> see :
>
> https://github.com/scikit-learn/scikit-learn/blob/master/sklearn/linear_model/tests/test_least_angle.py#L57
>
> Alex
>
> On Fri, Jun 29, 2012 at 1:40 AM, Charles-Pierre Astolfi
> wrote:
>> Btw, let
>>
>>>>> alphas, _,
terpolation
>
> Alex
>
> On Thu, Jun 28, 2012 at 6:42 PM, Olivier Grisel
> wrote:
>> 2012/6/28 Charles-Pierre Astolfi :
>>> Hi scikiters,
>>>
>>> I am trying to compute lasso/lars for several values of the
>>> regularization parameter.
&
Hi scikiters,
I am trying to compute lasso/lars for several values of the
regularization parameter.
A clean way to do this, is like this:
>>> alphas = [some long list of positive values]
>>> models = linear_model.lars_path(X, y, alphas=alphas)
Unfortunately, it seems to be very slow, compared to
Hi Alexandre,
>> 2. alphas should be sorted in ascending order? Or at least sorted?
>
> no. The path is fit starting from high alpha so alphas are returns in
> decreasing order.
Okay, thanks. I have an instance where this list of values is not
sorted. I'll submit a bug report.
--
Cp
---
Hi,
I'm using lars_path in order to get the regularization path of lasso.
I'd like to have some clarifications about the outputs:
alphas, _, coefs = lars_path(X, y, method='lasso')
Am I correct in assuming that:
1. alphas contains the hitting times of the parameter?
2. alphas should be sorted in
13 matches
Mail list logo
