Commit Graph

32 Commits

Author SHA1 Message Date
Lars Buitinck b8e1a6c357 ENH SGD Cython improvements
* nogil declarations where possible (but no with nogil blocks, since the
  performance benefit of those is not clear)
* use BLAS where appropriate
* replaced DOUBLE by double, INTEGER and int32 by int/np.intc, since that's
  the type used by scipy.sparse
* explicit checks that dataset dims don't exceed INT_MAX
* replaced an np.all by an explicit loop (npymath library for C99 isfinite)
2013-09-07 18:24:36 +02:00
Vlad Niculae e488649af4 ENH Add fast and stable logistic sigmoid to utils and RBM 2013-07-24 17:27:22 +02:00
Robert Layton 4e1e1ab772 Now that the folder has more than just mst in it, rename to sparsetools, which should help with referencing it. 2013-07-22 10:27:26 +02:00
Robert Layton 3ebd9ee775 Missed a spot 2013-07-22 10:27:25 +02:00
Robert Layton 67cdef8078 Moved mst to a subfolder and added a README file 2013-06-04 13:57:15 +02:00
Robert Layton 4c4f1cffb2 Minimal spanning tree backported from scipy 0.13
No tests in this commit
2013-06-04 13:57:15 +02:00
Lars Buitinck 68d2388f24 COSMIT translate lgamma replacement to C and clean it up 2013-01-31 01:22:08 +01:00
Vlad Niculae 452ca22102 ENH: use lgamma function from John D. Cook
Conflicts:
	sklearn/metrics/cluster/expected_mutual_info_fast.c
	sklearn/utils/lgamma.cpp
	sklearn/utils/lgamma.pyx
	sklearn/utils/setup.py
2013-01-13 14:03:45 +00:00
Lars Buitinck 76c66a0dab BUG Cholesky delete routines wouldn't compile on Solaris
Fixed that and cleaned it up. Should solve #1510.
2013-01-03 15:03:26 +01:00
Andreas Mueller 966e10c252 COSMIT PEP8 2012-12-22 16:48:17 +01:00
Andreas Mueller 754c2d9533 COSMIT pep8 2012-12-22 16:48:17 +01:00
Arnaud Joly a6ed58dc4d ENH flatten random_projection module + add sklearn.utils.random 2012-12-21 14:48:00 +01:00
Andreas Mueller 82a762f69d COSMIT pep8 2012-08-10 11:06:42 +01:00
GaelVaroquaux 446611dbcf MISC: build_utils: module rather than a subpackages 2012-07-26 11:45:20 +02:00
David Cournapeau d8e30f1ca8 REF: hack to be able to share distutils utilities.
We took the ugly trick from NumPy to separate build vs normal use of
sklearn imports.
2012-07-25 11:35:40 +01:00
Andreas Mueller c1c734af47 ENH put atlas checking in one place and load from there. 2012-07-20 23:45:26 +01:00
Andreas Mueller d4dc61b709 FIX trying to fix long-standing linker issue 2012-07-16 21:31:23 +01:00
Gael Varoquaux b32180db46 BUILD: add libm onlyon posix systems 2012-05-06 15:32:59 +02:00
Claire Revillet 690aaf1d1f - fix missing links to the C math libray 2012-05-06 15:32:59 +02:00
Peter Prettenhofer 7fcb2dbb3e moved WeightVector und SequentialDataset into seperate modules. 2012-03-07 20:09:05 +01:00
Peter Prettenhofer 3ea19f0d82 renamed largescale -> large_scale 2012-03-06 08:10:51 +01:00
Peter Prettenhofer add24160c3 added largescale ext module to setup.py 2012-02-15 21:30:36 +01:00
GaelVaroquaux ec20311e06 ENH: get the murmurhash to build properly 2012-01-16 08:11:50 +01:00
Olivier Grisel 6f6697896b first stab at trying to wrap MurmurHash3 2012-01-16 01:40:47 +01:00
Jake Vanderplas 7f53024ae9 remove unused imports 2011-12-23 09:27:40 -08:00
Mathieu Blondel 64adbcaddd Implement mean_variance_axis0. 2011-12-19 16:28:21 +01:00
Andreas Mueller 911eb2b6b6 COSMIT reindented for pep8 2011-11-20 21:18:41 +01:00
Lars Buitinck 48fda638a5 ENH incrementally build arrays in SVMlight loader to reduce memory usage
Memory use on covtype.binary [1] reduced by nearly 2/3, at the expense of
of a 40% slowdown.

[1] http://www.csie.ntu.edu.tw/~cjlin/libsvmtools/datasets/binary.html#covtype.binary
2011-11-17 19:43:00 +01:00
Olivier Grisel 78dd8b5626 batch trailing spaces cleanup 2011-09-06 23:12:26 +02:00
Fabian Pedregosa ddf4b72109 Move project directory from scikits.learn to sklearn 2011-09-02 12:06:57 +02:00
Fabian Pedregosa 68f27e3790 Revert "Move project directory from scikits.learn to sklearn"
This reverts commit fd0d3b879d.
2011-09-02 12:03:18 +02:00
Fabian Pedregosa fd0d3b879d Move project directory from scikits.learn to sklearn 2011-09-02 11:38:24 +02:00