c2aabf09bd
BUG rm LARS from linear_model.__init__
...
from sklearn.linear_model import * raised an exception.
2012-09-17 20:40:44 +02:00
5528334660
Robustify LARS. Fixes issue #487
...
Avoid division by zero when the y variable sums to zero.
2012-09-14 15:30:15 +02:00
63e99023b6
DOC fixed some sphinx errors, issues in docs....
2012-09-03 23:48:32 +01:00
b763ee23a7
DOC : better docstring for fit_path
2012-09-02 23:16:54 +02:00
28cecfeb6d
addressing for round of reviews
2012-09-02 23:16:54 +02:00
5af5b7f366
DOC : add useful comment in code
2012-09-02 23:16:54 +02:00
466479fe03
API : consistent use alpha_/alphas_ for alpha/alphas estimated by CV in linear models (issue #1041 )
2012-09-02 23:16:54 +02:00
105a67850b
update coef_path_ docstrings
2012-09-02 23:16:53 +02:00
d40f7a4246
Fix doctests
2012-09-02 23:16:53 +02:00
c1ef133651
Add possibility to ignore the path in Lars objects
2012-09-02 23:16:53 +02:00
eac812bca3
return_path option in lars_path
2012-09-02 23:16:53 +02:00
9c6422ce75
ENH : use multitarget Lars and LassoLars in dict_learning
2012-09-02 23:16:52 +02:00
519e204a95
FIX: add test, fix memory initialization bug
2012-09-02 23:16:51 +02:00
fee95e1428
Unnecessary copying
2012-09-02 23:16:51 +02:00
f74f5f09e0
ENH: move Gram precomputation outside of the loop
2012-09-02 23:16:51 +02:00
9dfdcc7096
First effort in multitarget lassolars
2012-09-02 23:16:51 +02:00
02a831a010
Remove LARS and LassoLARS
...
These classes were renamed and deprecated in version 0.9.
2012-09-01 20:02:13 +02:00
ca36d738ad
BUG: casting rule with recent numpy
...
Recent numpy doesn't accept inplace modification of integers by floats,
and its a good rule to avoid them.
2012-08-27 18:50:49 +03:00
a12bd7bc7c
ENH: understandable error message for X sparse
2012-08-10 11:23:00 +01:00
1864bca1df
COSMIT pep8
2012-07-09 22:27:44 +01:00
46bbddd2ae
Cosmit: PEP8 :P
2012-07-05 18:16:57 +02:00
a2170a5ae8
BUG: address ill-conditionned designs in Lars
...
When the design is too correlated, we end up with an ill-conditionned 'L'
matrix in the Lars that gives rise to a solve_choleski numerically
unstable.
Don't ask for a test: this is really hard to reproduce on toy datasets :)
2012-07-05 18:08:15 +02:00
fb714df433
Fix #904 .
2012-06-13 13:26:02 +09:00
294d4b6057
MISC added bigger "tiny" in lars_path. least_squares is float32.
2012-06-07 22:39:00 +02:00
3d32de1ecd
FIX : fix a division by zero in LARS #63
2012-06-06 08:05:06 +02:00
6592fdbee4
P3K: Fixing prints and dtypes
2012-05-11 11:24:27 +02:00
ca28fde531
fix for #824
2012-05-05 21:35:50 +02:00
741b047657
MISC: better verbosity in lars
2012-03-24 12:30:30 +01:00
1dbf4c3e7e
TEST: better tests of LassoCV and LassoLarsCV
2012-02-27 00:32:46 +01:00
a055f97f26
BUG: Same MSE scaling for LassoLarsCV and LassoCV
...
Fixes #619
2012-02-26 23:14:17 +01:00
56f59312d4
DOC: update dostring
2012-01-31 18:10:43 +01:00
a47773f2c9
ENH: numerical stability in LassoLarsCV
2012-01-31 18:10:43 +01:00
0dbd7adfb4
ENH : some doc + renaming in RandomizedLasso
2012-01-31 18:06:14 +01:00
fe5c85bec2
ENH: make LassoLarsIC more reproductible
2012-01-29 18:00:26 +01:00
6a973ddb4b
FIX: py3k's zip is not subscriptable
2011-12-29 09:54:08 +01:00
3a6e490e05
COSMIT pep8
2011-12-23 19:12:52 +01:00
ff517ddf32
DOC rm References sections in docstrings
...
According to @GaelVaroquaux, this breaks the NumPy docstring parser.
Moved references to Notes or See also sections, depending on context.
2011-12-21 18:05:22 +01:00
4683779e60
DOC: add alpha_ to attributes of LassoLarsIC
2011-12-21 01:17:43 +01:00
e9f432a3c4
DOC: make docstrings as good as I could
2011-12-20 01:13:49 +01:00
3aa5b12a15
DOC: add optimization objective to lasso and enet docstrings
2011-12-20 00:44:18 +01:00
25c93f4af6
COSMIT: pep8
2011-12-15 17:24:53 +01:00
05006c0adf
COSMIT use utils.deprecated as a class decorator
2011-12-11 00:55:39 +01:00
ff3f814a1b
COSMIT pep8 whitespace
2011-11-20 21:16:52 +01:00
b5429abd8d
Removed unused parameters in least_angle
2011-10-26 12:32:46 +02:00
c6aaef8091
MISC: More meaningful names for lapack functions in least_angle.
2011-10-26 12:25:28 +02:00
71d524396c
FIX replace np.atleast_2d with new utils.array2d
...
atleast_2d will pass through np.matrix
2011-10-25 15:16:31 +02:00
3bde2bf97b
ENH accept matrix input throughout
...
Simple solution: replace np.asanyarray with np.asarray throughout.
Also, use utils.as_float_array where appropriate.
2011-10-21 13:51:12 +02:00
8b2ca2af0f
Cosmetic changes in LARS
2011-10-17 12:09:45 +02:00
56d26f9fdd
rename overwrite_Foo params to copy_Foo (and inversed their meaning)
2011-09-25 16:52:29 +02:00
9955ed531c
DOC: fix typos
...
s/nn/n/ in dictionary and optional
2011-09-21 18:42:59 +02:00
Lars Buitinck