Commit Graph

71 Commits

Author SHA1 Message Date
Tom Dupré la Tour eed5fc5f0f Merge pull request #6395 from yenchenlin1994/make-dump_svmlight_file-support-sparse-y
[MRG+1] Make dump_svmlight_file support sparse y (fixes #6301)
2016-03-23 19:19:43 +01:00
YenChenLin 22d7cd5cd0 Make dump_svmlight_file support sparse y 2016-03-24 01:22:48 +08:00
Lars Buitinck 2e4aafd19c MAINT remove my email address
I'm getting too many individual support requests. People should
use the mailing list.
2016-03-04 10:43:41 +01:00
KamalakerDadi f2e35411fa Added more versions of 0.17 2015-11-04 00:02:52 +01:00
Chi Zhang 9d01a1c8d4 changed shape parameter 2015-09-02 14:18:59 -04:00
Chi Zhang 150267f9b1 added comments for y in dump_svmlight_file() method 2015-09-01 10:49:23 -04:00
Lars Buitinck 8932587159 FIX dump_svmlight_file multilabel arg: should be last
... in case someone somewhere used this function with positional
arguments. Spotted by @amueller.

Also pep8, listed in what's new.
2015-07-13 23:36:45 +02:00
Chih-Wei Chang 82ea224dfa Add multilabel support for dump_svmlight_file. 2015-07-12 13:51:27 +08:00
Raghav R V b3fbccca38 FIX various mismatch between docstring and signature params
DOC max_iterations -> max_iter. Make it consistent with kmeans

MAINT Replace the deprecated dx parameter with d in the docstrings

MAINT Deprecation warning for max_iterations parameter.
2015-01-16 11:54:57 -05:00
gpassino 969e677353 DOC fix kernels documentation inconsistencies
Fixes gh-3916, gh-2901.
2014-12-01 11:51:36 +01:00
Danny Sullivan 92a02ad13a adding example for memory wrapper 2014-10-23 16:51:21 +02:00
Lars Buitinck a7a05120e2 MAINT: handle frombuffer with empty 1st arg in utils.fixes 2014-09-22 17:23:51 +02:00
Andreas Mueller c235c3b566 ENH add allowed_sparse named argument for @ogrisel 2014-07-20 15:31:28 +02:00
Andreas Mueller 6e2a83b4e1 remove check_arrays stuff and old input validation 2014-07-20 13:31:45 +02:00
Lars Buitinck bb743ddb2a ENH speed up SVMlight loader using Cython's array support
Loading the binary covtype dataset from
http://www.csie.ntu.edu.tw/~cjlin/libsvmtools/datasets/binary.html
now takes 1/3 less wall clock time, which includes disk I/O and
decompression (bzip2).

Got rid of ArrayBuilder in the process.
2014-01-05 22:10:02 +01:00
Lars Buitinck 119cc2407f BUG SVMlight loader should check whether n_features is big enough
scipy.sparse won't complain when it gets a dimensionality that is
too large, causing crashes in estimators.

Fixes #1630.
2013-11-25 19:09:43 +01:00
Lars Buitinck ad15405f3c ENH 10x speedup in dump_svmlight_format
As measured by dumping

X = rand(10000, 4000, density=.0125)
y = np.random.randint(0, 5, X.shape[0])

to /dev/null.
2013-09-20 12:13:00 +02:00
Jaques Grobler 8861833ec9 DOC spellfixes 2013-06-29 16:18:23 +03:00
Mathieu Blondel 64760f25e7 Import gzip and bz2 only if necessary. 2013-05-09 17:40:33 +09:00
Robert Layton dacfd8bd5d DOC: Replaced all BSD style licenses with "BSD 3 clause"
Replaced all BSD style licenses with "BSD 3 clause"
Not checked yet!

Removed duplicate "3 clause, 3 clause"

Removed trailing period if exists

Fixed some missed licences, still about 50 to do, but those can be automated

Think I got the last of them.

Apparently me and sed have different ideas of regex.

Found a few more
2013-04-30 08:34:46 +02:00
Seamus Abshere ff0e88213f ENH reduce size of files produced by dump_svmlight_file 2013-04-16 12:14:58 +02:00
Andreas Mueller a1d4d184fe COSMIT pep8 2013-04-02 13:30:55 +02:00
Steven De Gryze 5993ef5082 PY3: use b() convenience function for string literals
Had to use an exception in lfw.py; six.u didn't work
2013-03-19 17:57:02 +01:00
Raul Garreta 71d1a7c135 PY3: used six.u to fix unicode variables in svmlight 2013-03-19 17:57:02 +01:00
Mathieu Blondel 50e22dd677 Update load_svmlight_file docstring. 2013-03-06 20:30:27 +09:00
Olivier Grisel 1967a0b323 P3K use six to have a python 2 & 3 compatible code base 2013-02-09 18:08:44 +01:00
Lars Buitinck 57b86fc434 P3K use six.string_types and six.PY3 2013-02-03 20:12:42 +01:00
Lars Buitinck c59af39337 BUG hotfix for issue #1501: sort indices in SVMlight i/o
CSR matrices are strange beasts.
2013-01-17 00:40:13 +01:00
Lars Buitinck 49e9391168 BUG don't write comments in SVMlight dumper by default
Fixes part of Issue #1501
2013-01-16 17:19:23 +01:00
Andreas Mueller 754c2d9533 COSMIT pep8 2012-12-22 16:48:17 +01:00
Lars Buitinck 262cbdf8ad DOC clarify zero_based's implications in SVMlight loader 2012-10-11 11:26:22 +02:00
Andreas Mueller 57001002bb COSMIT pep8 2012-10-10 09:38:10 +02:00
Fabian Pedregosa 04b5e98400 Add support for query_id in dump_svmlight_file
Added a test and docstring
2012-09-29 10:19:07 +02:00
Fabian Pedregosa 6c1e0ed858 FIX: typo in ValueError message.
It was printing the full array when it should be printing the shape
2012-09-29 09:46:06 +02:00
Fabian Pedregosa 398fa9d136 FIX: query_id parameter and other cosmetic changes 2012-09-28 11:24:28 +02:00
Fabian Pedregosa 6fd39fbf12 Add support for preference contraints in svmlight format.
This feature is explained in http://svmlight.joachims.org/ and adds
the ability to control which pairs are considered when dealing with
pairwise loss functions (e.g. for SVMrank).

In order to preserve API compatibility, I return this information (if
present) as a second column to the y vector.
2012-09-26 17:48:13 +02:00
Lars Buitinck 35b24cac4c ENH allow user-specified comment in SVMlight dumper 2012-07-23 10:46:18 +02:00
Lars Buitinck e1ba110829 ENH make generated SVMlight files self-describing in a comment
Comment is *not* used to parse the files, since that would break
compatibility with other tools; it's just there for the user.
2012-07-21 14:32:29 +02:00
Lars Buitinck 2530791e2e ENH better input validation for dump_svmlight_file
Also, do validation *before* opening any files.
2012-07-21 14:32:29 +02:00
Olivier Grisel 0240e3d85b work on smmlight serualizaer to preserve double precision values 2012-07-10 01:05:50 +02:00
Olivier Grisel 51661e7bb3 FIX #937: preserve double precision values in svmlight serializer 2012-07-07 20:14:58 +02:00
Lars Buitinck 669bbf2de1 BUG do not close fd passed by user in SVMlight loader
+ document the behavior of load_svmlight_file better
2012-06-13 16:31:39 +02:00
Lars Buitinck ddb7b4df81 ENH always support file descriptors in SVMlight loader
Previously only supported in Python 3.x
2012-06-11 14:11:50 +02:00
Lars Buitinck 4eedded8db ENH support opening compressed files in SVMlight reader 2012-06-04 10:50:51 +02:00
Lars Buitinck d38f4d0e91 DOC + PEP8 SVMlight loader 2012-04-16 10:09:57 +02:00
Lars Buitinck 96af5ae934 TEST + FIX zero_based="auto" behavior in SVMlight loader 2012-04-15 16:50:51 +02:00
Lars Buitinck 7b4842d1db BUG close files in time in SVMlight loader (with statement) 2012-04-15 16:39:29 +02:00
Lars Buitinck 2ddf1a7b6d COSMIT pep8 SVMlight loader 2012-04-15 16:33:22 +02:00
Lars Buitinck c2eeca76ae COSMIT refactor SVMlight loader 2012-04-14 23:03:53 +02:00
Lars Buitinck 6d6c9a8937 ENH zero_based="auto" support + better n_features=None in load_svmlight_files 2012-04-10 18:20:28 +02:00