8ffee3f4f4
Wrapping line in naive_bayes.rst and specifying bin_seeding argument as optional
2013-10-16 15:52:50 +05:30
3a770a2dbf
Adding doc for bin_seeding parameter in cluster.MeanShift
2013-10-15 00:13:44 +05:30
47100454b4
Making the results NaiveBayes example more explicit
2013-10-14 14:28:24 +05:30
551625fad7
MAINT simplify f_oneway in feature selection
...
reduce(lambda x, y: x + y, s) == sum(s), always.
2013-10-14 00:15:31 +02:00
9e10ec017b
COSMIT use norm function in feature selection
2013-10-13 23:20:39 +02:00
551b6c2f06
ENH honor Y_norm_squared when X=Y in euclidean_distances
...
Also a micro-optimization to row_norms.
2013-10-13 18:54:21 +02:00
a2d95d3e83
Merge branch 'refactor-squared-norms'
2013-10-13 17:31:17 +02:00
761dde684a
COSMIT remove dead code in k-means
...
Private functions _k_init and _kmeans_single were doing too much work.
Should improve the relative coverage.
2013-10-13 17:30:47 +02:00
a2b13bd438
MAINT use subprocess.call, not os.system
...
The former is safer and faster.
2013-10-13 16:39:44 +02:00
a6b891306a
FIX: clean-build target does no longer exist
2013-10-13 16:34:13 +02:00
55a2f29ee0
FIX fit followed by partial_fit in multiclass SGD
...
Fixes #2496 and includes a non-regression test. Since the introduction
of fast_dot (and NumPy 1.6), dot products between arrays of various
memory layouts have become fast enough to no longer require storing coef_
in Fortran order. The only noticeably slower operation is multiplying a
CSR matrix with a Fortran-ordered array, but then building the CSR matrix
in the first place is more likely to be the bottleneck.
2013-10-13 15:55:02 +02:00
b7f482bd3c
ENH use row_norms in KDE code
...
There may be other places where it is useful.
2013-10-12 20:27:54 +02:00
24af95a211
ENH refactor squared-norms computation to extmath
...
k-means can now use the memory-efficient dense implementation from
metrics.pairwise, while pairwise can use the fast sparse implementation
from k-means.
2013-10-12 20:27:51 +02:00
edf747e6a2
DOC docstring for extmath.norm
2013-10-12 20:17:33 +02:00
809378f259
FIX parallel FeatureUnion.fit and fit_transform
...
It didn't work previously because transformers were fit in another
process, and information they learned was not transferred back.
2013-10-12 20:15:24 +02:00
c1177990f2
MAINT: use python setup.py clean in Makefile
2013-10-12 20:12:39 +02:00
ef108b0b21
MAINT add authors to validation.py and pairwise.py
2013-10-12 20:09:41 +02:00
f01e215ec8
Merge pull request #2502 from ericjster/plot_dbscan
...
ENH: Call plot twice per class label rather than for every point.
2013-10-12 09:34:16 -07:00
5df97f6813
Merge branch 'plot_dbscan' of https://github.com/ericjster/scikit-learn into plot_dbscan
...
Fix formatting to be PEP8.
2013-10-10 01:22:28 -07:00
0974fb6d3c
ENH: Call plot twice per class label rather than for every point.
2013-10-10 00:37:31 -07:00
e3211e1c20
Cosmit: move log_loss.
2013-10-10 11:57:48 +09:00
d61ffe8323
Merge pull request #2487 from jaquesgrobler/DOC_speed_up_frontpage
...
another speedup attempt for index.html:
2013-10-09 06:51:30 -07:00
29916aeac3
ENH: Call plot twice per class label rather than for every point.
2013-10-09 02:49:32 -07:00
097874c6d1
Merge pull request #2495 from jaquesgrobler/ENH_coverage_travis
...
add coverage test for travis
2013-10-09 02:32:16 -07:00
2c05daf4e6
Merge pull request #2500 from ndawe/r2
...
[MRG] AdaBoostRegressor: fix redundant recalculation of error_vect.max()
2013-10-08 23:36:26 -07:00
2988a23089
AdaBoostRegressor: fix redundant recalculation of error_vect.max()
2013-10-08 20:15:41 -07:00
a3995ecf73
Merge pull request #2497 from dengemann/extend_fast_dot
...
FIX: support older scipy API for fast_dot
2013-10-08 14:03:38 -07:00
c1df89edf2
... add the fix to import check
2013-10-08 23:01:48 +02:00
add591cbf7
FIX scipy API
2013-10-08 22:54:14 +02:00
015113a59e
remove whitespace, fix pip install
2013-10-08 16:33:01 +02:00
8f9f65850d
coverall added
2013-10-08 15:08:13 +02:00
26dcdfa105
add coverage test for travis
2013-10-08 10:31:02 +02:00
6f3a8057fa
Merge pull request #2489 from samuela/patch-1
...
Fix typo in the Gaussian PDF in Naive Bayes docs
2013-10-05 06:56:11 -07:00
c326be1e78
Fix typo in the Gaussian PDF
...
Hopefully, this same issue doesn't exist in the code.
I would also feel more comfortable if mu_y and sigma_y were mu_iy and sigma_iy, respectively, seeing as there are separate distributions per feature, per class.
2013-10-04 17:54:05 -04:00
2cae5d6bf6
logo lossless compressions
2013-10-04 14:53:43 +02:00
09d1ce74d7
switch species_kde to thumbnail
2013-10-04 14:13:13 +02:00
eade1b73e1
add lossless image compression for docbuilding machine
2013-10-04 13:33:33 +02:00
d0c3395720
ENH: small optimization in PresortBestSplitter
2013-10-04 13:30:53 +02:00
7f6e398d1e
remove paren
2013-10-04 10:58:03 +02:00
e95533d172
cleanup changes
2013-10-03 16:48:00 +02:00
bb2d007119
another speedup attempt for index.html: carousel images to use thumbnails instead, 1 blocking js removed for page (sidebar.js)
2013-10-03 15:02:39 +02:00
1d0a3fd87e
DOC error in feature_extraction.text docstrings
...
Spotted by @madisonmay. Fixes #2486 .
2013-10-02 19:13:21 +02:00
baad8de295
COSMIT pyflakes feature_extraction.text tests
2013-10-02 19:10:37 +02:00
94d66a0e38
FIX max_features in CountVectorizer
...
Fixes #2443 : max_features would be selected based on document frequency
rather than term frequency.
2013-10-02 19:10:37 +02:00
a2a4257bc6
ENH optimize NMF inner loop
...
The call to np.maximum was taking 20% of the time.
Multiplying by Hn > 0 is about twice as fast.
This reverts commit 6ef08cfac2
2013-10-01 18:09:58 +02:00
81950ba611
ENH sparse matrix support in pairwise + optimizations
2013-09-30 19:22:53 +02:00
0d2c8c17e8
DOC improve biclustering docstrings
...
I had to figure out by experimentation what the methods were doing.
2013-09-30 12:33:24 +02:00
56c06310da
update what's new
2013-09-30 09:08:10 +02:00
af4b940701
ENH : simplify input checking in GP
2013-09-30 09:06:00 +02:00
f244a285a8
ENH : Gaussian process for arbitrary-dimensional output spaces by @JohntheBear
2013-09-30 09:06:00 +02:00
Ankit Agrawal