-
Notifications
You must be signed in to change notification settings - Fork 2
/
Copy pathwhats_new.html
3019 lines (2980 loc) · 305 KB
/
whats_new.html
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"
"http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">
<html xmlns="http://www.w3.org/1999/xhtml">
<head>
<meta http-equiv="Content-Type" content="text/html; charset=utf-8" />
<title>0.16 — scikit-learn 0.15.2 documentation</title>
<!-- htmltitle is before nature.css - we use this hack to load bootstrap first -->
<meta name="viewport" content="width=device-width, initial-scale=1.0" />
<link rel="stylesheet" href="_static/css/bootstrap.min.css" media="screen" />
<link rel="stylesheet" href="_static/css/bootstrap-responsive.css"/>
<link rel="stylesheet" href="_static/nature.css" type="text/css" />
<link rel="stylesheet" href="_static/pygments.css" type="text/css" />
<link rel="stylesheet" href="_static/gallery.css" type="text/css" />
<script type="text/javascript">
var DOCUMENTATION_OPTIONS = {
URL_ROOT: './',
VERSION: '0.15.2',
COLLAPSE_INDEX: false,
FILE_SUFFIX: '.html',
HAS_SOURCE: true
};
</script>
<script type="text/javascript" src="_static/jquery.js"></script>
<script type="text/javascript" src="_static/underscore.js"></script>
<script type="text/javascript" src="_static/doctools.js"></script>
<script type="text/javascript" src="_static/js/copybutton.js"></script>
<link rel="shortcut icon" href="_static/favicon.ico"/>
<link rel="author" title="About these documents" href="about.html" />
<link rel="top" title="scikit-learn 0.15.2 documentation" href="index.html" />
<link rel="next" title="External Resources, Videos and Talks" href="presentations.html" />
<link rel="prev" title="Support" href="support.html" />
<script type="text/javascript" src="_static/sidebar.js"></script>
<meta name="viewport" content="width=device-width, initial-scale=1.0" />
<script src="_static/js/bootstrap.min.js" type="text/javascript"></script>
<link rel="canonical" href="http://scikit-learn.org/stable/whats_new.html" />
<script type="text/javascript">
$("div.buttonNext, div.buttonPrevious").hover(
function () {
$(this).css('background-color', '#FF9C34');
},
function () {
$(this).css('background-color', '#A7D6E2');
}
);
var bodywrapper = $('.bodywrapper');
var sidebarbutton = $('#sidebarbutton');
sidebarbutton.css({'height': '900px'});
</script>
</head>
<body>
<div class="header-wrapper">
<div class="header">
<p class="logo"><a href="index.html">
<img src="_static/scikit-learn-logo-small.png" alt="Logo"/>
</a>
</p><div class="navbar">
<ul>
<li><a href="../dev/index.html">Home</a></li>
<li><a href="../dev/install.html">Installation</a></li>
<li class="btn-li"><div class="btn-group">
<a href="documentation.html">Documentation</a>
<a class="btn dropdown-toggle" data-toggle="dropdown">
<span class="caret"></span>
</a>
<ul class="dropdown-menu">
<li class="link-title">Scikit-learn 0.16 (development)</li>
<li><a href="tutorial/index.html">Tutorials</a></li>
<li><a href="user_guide.html">User guide</a></li>
<li><a href="modules/classes.html">API</a></li>
<li><a href="faq.html">FAQ</a></li>
<li class="divider"></li>
<li><a href="http://scikit-learn.org/stable/documentation.html">Stable</a></li>
<li><a href="http://scikit-learn.org/0.14/">Scikit-learn 0.14</a></li>
<li><a href="http://scikit-learn.org/0.13/">Scikit-learn 0.13</a></li>
<li><a href="http://scikit-learn.org/0.12/">Scikit-learn 0.12</a></li>
<li><a href="http://scikit-learn.org/0.11/">Scikit-learn 0.11</a></li>
<li><a href="documentation.html">More versions...</a></li>
</ul>
</div>
</li>
<li><a href="auto_examples/index.html">Examples</a></li>
</ul>
<div class="search_form">
<div id="cse" style="width: 100%;"></div>
</div>
</div> <!-- end navbar --></div>
</div>
<!-- Github "fork me" ribbon -->
<a href="https://github.com/scikit-learn/scikit-learn">
<img class="fork-me"
style="position: absolute; top: 0; right: 0; border: 0;"
src="_static/img/forkme.png"
alt="Fork me on GitHub" />
</a>
<div class="content-wrapper">
<div class="sphinxsidebar">
<div class="sphinxsidebarwrapper">
<div class="rel rellarge">
<!-- rellinks[1:] is an ugly hack to avoid link to module
index -->
<div class="rellink">
<a href="support.html"
accesskey="P">Previous
<br/>
<span class="smallrellink">
Support
</span>
<span class="hiddenrellink">
Support
</span>
</a>
</div>
<div class="spacer">
</div>
<div class="rellink">
<a href="presentations.html"
accesskey="N">Next
<br/>
<span class="smallrellink">
External Resourc...
</span>
<span class="hiddenrellink">
External Resources, Videos and Talks
</span>
</a>
</div>
<!-- Ad a link to the 'up' page -->
</div>
<p class="doc-version">This documentation is for scikit-learn <strong>version 0.15.2</strong> — <a href="http://scikit-learn.org/stable/support.html#documentation-resources">Other versions</a></p>
<p class="citing">If you use the software, please consider <a href="about.html#citing-scikit-learn">citing scikit-learn</a>.</p>
<ul>
<li><a class="reference internal" href="#">0.16</a><ul>
<li><a class="reference internal" href="#changelog">Changelog</a><ul>
<li><a class="reference internal" href="#new-features">New features</a></li>
<li><a class="reference internal" href="#enhancements">Enhancements</a></li>
<li><a class="reference internal" href="#documentation-improvements">Documentation improvements</a></li>
<li><a class="reference internal" href="#bug-fixes">Bug fixes</a></li>
</ul>
</li>
<li><a class="reference internal" href="#api-changes-summary">API changes summary</a></li>
</ul>
</li>
<li><a class="reference internal" href="#changes-0-15-2">0.15.2</a><ul>
<li><a class="reference internal" href="#id3">Bug fixes</a></li>
</ul>
</li>
<li><a class="reference internal" href="#changes-0-15-1">0.15.1</a><ul>
<li><a class="reference internal" href="#id5">Bug fixes</a></li>
</ul>
</li>
<li><a class="reference internal" href="#changes-0-15">0.15</a><ul>
<li><a class="reference internal" href="#highlights">Highlights</a></li>
<li><a class="reference internal" href="#id7">Changelog</a><ul>
<li><a class="reference internal" href="#id8">New features</a></li>
<li><a class="reference internal" href="#id9">Enhancements</a></li>
<li><a class="reference internal" href="#id10">Documentation improvements</a></li>
<li><a class="reference internal" href="#id11">Bug fixes</a></li>
</ul>
</li>
<li><a class="reference internal" href="#id12">API changes summary</a></li>
<li><a class="reference internal" href="#people">People</a></li>
</ul>
</li>
<li><a class="reference internal" href="#changes-0-14">0.14</a><ul>
<li><a class="reference internal" href="#id14">Changelog</a></li>
<li><a class="reference internal" href="#id15">API changes summary</a></li>
<li><a class="reference internal" href="#id16">People</a></li>
</ul>
</li>
<li><a class="reference internal" href="#changes-0-13-1">0.13.1</a><ul>
<li><a class="reference internal" href="#id18">Changelog</a></li>
<li><a class="reference internal" href="#id19">People</a></li>
</ul>
</li>
<li><a class="reference internal" href="#changes-0-13">0.13</a><ul>
<li><a class="reference internal" href="#new-estimator-classes">New Estimator Classes</a></li>
<li><a class="reference internal" href="#id21">Changelog</a></li>
<li><a class="reference internal" href="#id22">API changes summary</a></li>
<li><a class="reference internal" href="#id23">People</a></li>
</ul>
</li>
<li><a class="reference internal" href="#changes-0-12-1">0.12.1</a><ul>
<li><a class="reference internal" href="#id25">Changelog</a></li>
<li><a class="reference internal" href="#id26">People</a></li>
</ul>
</li>
<li><a class="reference internal" href="#changes-0-12">0.12</a><ul>
<li><a class="reference internal" href="#id28">Changelog</a></li>
<li><a class="reference internal" href="#id29">API changes summary</a></li>
<li><a class="reference internal" href="#id30">People</a></li>
</ul>
</li>
<li><a class="reference internal" href="#changes-0-11">0.11</a><ul>
<li><a class="reference internal" href="#id32">Changelog</a><ul>
<li><a class="reference internal" href="#id33">Highlights</a></li>
<li><a class="reference internal" href="#other-changes">Other changes</a></li>
</ul>
</li>
<li><a class="reference internal" href="#id34">API changes summary</a></li>
<li><a class="reference internal" href="#id35">People</a></li>
</ul>
</li>
<li><a class="reference internal" href="#changes-0-10">0.10</a><ul>
<li><a class="reference internal" href="#id37">Changelog</a></li>
<li><a class="reference internal" href="#id38">API changes summary</a></li>
<li><a class="reference internal" href="#id39">People</a></li>
</ul>
</li>
<li><a class="reference internal" href="#changes-0-9">0.9</a><ul>
<li><a class="reference internal" href="#id41">Changelog</a></li>
<li><a class="reference internal" href="#id42">API changes summary</a></li>
<li><a class="reference internal" href="#id43">People</a></li>
</ul>
</li>
<li><a class="reference internal" href="#changes-0-8">0.8</a><ul>
<li><a class="reference internal" href="#id45">Changelog</a></li>
<li><a class="reference internal" href="#id46">People</a></li>
</ul>
</li>
<li><a class="reference internal" href="#changes-0-7">0.7</a><ul>
<li><a class="reference internal" href="#id48">Changelog</a></li>
<li><a class="reference internal" href="#id49">People</a></li>
</ul>
</li>
<li><a class="reference internal" href="#changes-0-6">0.6</a><ul>
<li><a class="reference internal" href="#id51">Changelog</a></li>
<li><a class="reference internal" href="#id52">People</a></li>
</ul>
</li>
<li><a class="reference internal" href="#changes-0-5">0.5</a><ul>
<li><a class="reference internal" href="#id54">Changelog</a></li>
<li><a class="reference internal" href="#new-classes">New classes</a></li>
<li><a class="reference internal" href="#documentation">Documentation</a></li>
<li><a class="reference internal" href="#fixes">Fixes</a></li>
<li><a class="reference internal" href="#examples">Examples</a></li>
<li><a class="reference internal" href="#external-dependencies">External dependencies</a></li>
<li><a class="reference internal" href="#removed-modules">Removed modules</a></li>
<li><a class="reference internal" href="#misc">Misc</a></li>
<li><a class="reference internal" href="#authors">Authors</a></li>
</ul>
</li>
<li><a class="reference internal" href="#id55">0.4</a><ul>
<li><a class="reference internal" href="#id56">Changelog</a></li>
<li><a class="reference internal" href="#id57">Authors</a></li>
</ul>
</li>
<li><a class="reference internal" href="#earlier-versions">Earlier versions</a></li>
</ul>
</div>
</div>
<div class="content">
<div class="documentwrapper">
<div class="bodywrapper">
<div class="body">
<div class="section" id="changes-0-16">
<span id="id1"></span><h1>0.16<a class="headerlink" href="#changes-0-16" title="Permalink to this headline">¶</a></h1>
<div class="section" id="changelog">
<h2>Changelog<a class="headerlink" href="#changelog" title="Permalink to this headline">¶</a></h2>
<div class="section" id="new-features">
<h3>New features<a class="headerlink" href="#new-features" title="Permalink to this headline">¶</a></h3>
<blockquote>
<div><ul class="simple">
<li>The new <tt class="xref py py-class docutils literal"><span class="pre">neighbors.LSHForest</span></tt> implements locality-sensitive hashing
for approximate nearest neighbors search. By <a class="reference external" href="https://github.com/maheshakya">Maheshakya Wijewardena</a>.</li>
<li>Added <tt class="xref py py-class docutils literal"><span class="pre">svm.LinearSVR</span></tt>. This class uses the liblinear implementation
of Support Vector Regression which is much faster for large
sample sizes than <a class="reference internal" href="modules/generated/sklearn.svm.SVR.html#sklearn.svm.SVR" title="sklearn.svm.SVR"><tt class="xref py py-class docutils literal"><span class="pre">svm.SVR</span></tt></a> with linear kernel. By
<a class="reference external" href="http://fseoane.net/blog/">Fabian Pedregosa</a> and Qiang Luo.</li>
<li>Incremental fit for <a class="reference internal" href="modules/generated/sklearn.naive_bayes.GaussianNB.html#sklearn.naive_bayes.GaussianNB" title="sklearn.naive_bayes.GaussianNB"><tt class="xref py py-class docutils literal"><span class="pre">GaussianNB</span></tt></a>.</li>
<li>Added <tt class="docutils literal"><span class="pre">sample_weight</span></tt> support to <a class="reference internal" href="modules/generated/sklearn.dummy.DummyClassifier.html#sklearn.dummy.DummyClassifier" title="sklearn.dummy.DummyClassifier"><tt class="xref py py-class docutils literal"><span class="pre">dummy.DummyClassifier</span></tt></a> and
<a class="reference internal" href="modules/generated/sklearn.dummy.DummyRegressor.html#sklearn.dummy.DummyRegressor" title="sklearn.dummy.DummyRegressor"><tt class="xref py py-class docutils literal"><span class="pre">dummy.DummyRegressor</span></tt></a>. By <a class="reference external" href="http://www.ajoly.org">Arnaud Joly</a>.</li>
<li>Added the <tt class="xref py py-func docutils literal"><span class="pre">metrics.label_ranking_average_precision_score</span></tt> metrics.
By <a class="reference external" href="http://www.ajoly.org">Arnaud Joly</a>.</li>
<li>Add the <tt class="xref py py-func docutils literal"><span class="pre">metrics.coverage_error</span></tt> metrics. By <a class="reference external" href="http://www.ajoly.org">Arnaud Joly</a>.</li>
<li>Added <tt class="xref py py-class docutils literal"><span class="pre">linear_model.LogisticRegressionCV</span></tt>. By
<a class="reference external" href="https://manojbits.wordpress.com">Manoj Kumar</a>, <a class="reference external" href="http://fseoane.net/blog/">Fabian Pedregosa</a>, <a class="reference external" href="http://gael-varoquaux.info">Gael Varoquaux</a>
and <a class="reference external" href="http://alexandre.gramfort.net">Alexandre Gramfort</a>.</li>
<li>Added <tt class="docutils literal"><span class="pre">warm_start</span></tt> constructor parameter to make it possible for any
trained forest model to grow additional trees incrementally. By
<a class="reference external" href="https://github.com/ldirer">Laurent Direr</a>.</li>
<li>Added <tt class="docutils literal"><span class="pre">sample_weight</span></tt> support to <a class="reference internal" href="modules/generated/sklearn.ensemble.GradientBoostingClassifier.html#sklearn.ensemble.GradientBoostingClassifier" title="sklearn.ensemble.GradientBoostingClassifier"><tt class="xref py py-class docutils literal"><span class="pre">ensemble.GradientBoostingClassifier</span></tt></a> and
<a class="reference internal" href="modules/generated/sklearn.ensemble.GradientBoostingRegressor.html#sklearn.ensemble.GradientBoostingRegressor" title="sklearn.ensemble.GradientBoostingRegressor"><tt class="xref py py-class docutils literal"><span class="pre">ensemble.GradientBoostingRegressor</span></tt></a>. By <a class="reference external" href="http://sites.google.com/site/peterprettenhofer/">Peter Prettenhofer</a>.</li>
<li>Added <tt class="xref py py-class docutils literal"><span class="pre">decomposition.IncrementalPCA</span></tt>, an implementation of the PCA
algorithm that supports out-of-core learning with a <tt class="docutils literal"><span class="pre">partial_fit</span></tt>
method. By <a class="reference external" href="http://kastnerkyle.github.io">Kyle Kastner</a>.</li>
<li>Averaged SGD for <a class="reference internal" href="modules/generated/sklearn.linear_model.SGDClassifier.html#sklearn.linear_model.SGDClassifier" title="sklearn.linear_model.SGDClassifier"><tt class="xref py py-class docutils literal"><span class="pre">SGDClassifier</span></tt></a>
and <a class="reference internal" href="modules/generated/sklearn.linear_model.SGDRegressor.html#sklearn.linear_model.SGDRegressor" title="sklearn.linear_model.SGDRegressor"><tt class="xref py py-class docutils literal"><span class="pre">SGDRegressor</span></tt></a> By
<a class="reference external" href="https://github.com/dsullivan7">Danny Sullivan</a>.</li>
<li>Added <tt class="xref py py-func docutils literal"><span class="pre">cross_val_predict</span></tt>
function which computes cross-validated estimates. By <a class="reference external" href="http://luispedro.org">Luis Pedro Coelho</a></li>
<li>Added <tt class="xref py py-class docutils literal"><span class="pre">linear_model.TheilSenRegressor</span></tt>, a robust
generalized-median-based estimator. By <a class="reference external" href="https://github.com/FlorianWilhelm">Florian Wilhelm</a>.</li>
<li>Added <tt class="xref py py-func docutils literal"><span class="pre">metrics.median_absolute_error</span></tt>, a robust metric.
By <a class="reference external" href="http://gael-varoquaux.info">Gael Varoquaux</a> and <a class="reference external" href="https://github.com/FlorianWilhelm">Florian Wilhelm</a>.</li>
<li>Add <tt class="xref py py-class docutils literal"><span class="pre">cluster.Birch</span></tt>, an online clustering algorithm. By
<a class="reference external" href="https://manojbits.wordpress.com">Manoj Kumar</a>, <a class="reference external" href="http://alexandre.gramfort.net">Alexandre Gramfort</a> and <a class="reference external" href="http://joelnothman.com">Joel Nothman</a>.</li>
<li>Added shrinkage support to <a class="reference internal" href="modules/generated/sklearn.lda.LDA.html#sklearn.lda.LDA" title="sklearn.lda.LDA"><tt class="xref py py-class docutils literal"><span class="pre">lda.LDA</span></tt></a> using two new solvers. By
<a class="reference external" href="https://github.com/cle1109">Clemens Brunner</a> and <a class="reference external" href="https://github.com/kazemakase">Martin Billinger</a>.</li>
<li>Added <tt class="xref py py-class docutils literal"><span class="pre">kernel_ridge.KernelRidge</span></tt>, an implementation of
kernelized ridge regression.
By <a class="reference external" href="http://www.mblondel.org">Mathieu Blondel</a> and <a class="reference external" href="https://jmetzen.github.io/">Jan Hendrik Metzen</a>.</li>
<li>All solvers in <a class="reference internal" href="modules/generated/sklearn.linear_model.Ridge.html#sklearn.linear_model.Ridge" title="sklearn.linear_model.Ridge"><tt class="xref py py-class docutils literal"><span class="pre">linear_model.Ridge</span></tt></a> now support <cite>sample_weight</cite>.
By <a class="reference external" href="http://www.mblondel.org">Mathieu Blondel</a>.</li>
<li>Added <tt class="xref py py-class docutils literal"><span class="pre">cross_validation.PredefinedSplit</span></tt> cross-validation
for fixed user-provided cross-validation folds.
By <a class="reference external" href="https://github.com/untom">untom</a>.</li>
</ul>
</div></blockquote>
</div>
<div class="section" id="enhancements">
<h3>Enhancements<a class="headerlink" href="#enhancements" title="Permalink to this headline">¶</a></h3>
<blockquote>
<div><ul class="simple">
<li>Add option <tt class="docutils literal"><span class="pre">return_distance</span></tt> in <tt class="xref py py-func docutils literal"><span class="pre">hierarchical.ward_tree</span></tt>
to return distances between nodes for both structured and unstructured
versions of the algorithm. By <a class="reference external" href="http://www.mvdoc.me">Matteo Visconti di Oleggio Castello</a>.
The same option was added in <tt class="xref py py-func docutils literal"><span class="pre">hierarchical.linkage_tree</span></tt>.
By <a class="reference external" href="https://manojbits.wordpress.com">Manoj Kumar</a></li>
<li>Add support for sample weights in scorer objects. Metrics with sample
weight support will automatically benefit from it. By <a class="reference external" href="http://noel.dawe.me">Noel Dawe</a> and
<a class="reference external" href="http://vene.ro">Vlad Niculae</a>.</li>
<li>Added <tt class="docutils literal"><span class="pre">newton-cg</span></tt> and <cite>lbfgs</cite> solver support in
<a class="reference internal" href="modules/generated/sklearn.linear_model.LogisticRegression.html#sklearn.linear_model.LogisticRegression" title="sklearn.linear_model.LogisticRegression"><tt class="xref py py-class docutils literal"><span class="pre">linear_model.LogisticRegression</span></tt></a>. By <a class="reference external" href="https://manojbits.wordpress.com">Manoj Kumar</a>.</li>
<li>Add <tt class="docutils literal"><span class="pre">selection="random"</span></tt> parameter to implement stochastic coordinate
descent for <a class="reference internal" href="modules/generated/sklearn.linear_model.Lasso.html#sklearn.linear_model.Lasso" title="sklearn.linear_model.Lasso"><tt class="xref py py-class docutils literal"><span class="pre">linear_model.Lasso</span></tt></a>, <a class="reference internal" href="modules/generated/sklearn.linear_model.ElasticNet.html#sklearn.linear_model.ElasticNet" title="sklearn.linear_model.ElasticNet"><tt class="xref py py-class docutils literal"><span class="pre">linear_model.ElasticNet</span></tt></a>
and related. By <a class="reference external" href="https://manojbits.wordpress.com">Manoj Kumar</a>.</li>
<li>Add <tt class="docutils literal"><span class="pre">sample_weight</span></tt> parameter to <cite>metrics.jaccard_similarity_score</cite> and
<cite>metrics.log_loss</cite>. By <a class="reference external" href="http://jatinshah.org/">Jatin Shah</a>.</li>
<li>Support sparse multilabel indicator representation in
<a class="reference internal" href="modules/generated/sklearn.preprocessing.LabelBinarizer.html#sklearn.preprocessing.LabelBinarizer" title="sklearn.preprocessing.LabelBinarizer"><tt class="xref py py-class docutils literal"><span class="pre">preprocessing.LabelBinarizer</span></tt></a> and
<a class="reference internal" href="modules/generated/sklearn.multiclass.OneVsRestClassifier.html#sklearn.multiclass.OneVsRestClassifier" title="sklearn.multiclass.OneVsRestClassifier"><tt class="xref py py-class docutils literal"><span class="pre">multiclass.OneVsRestClassifier</span></tt></a> (by <a class="reference external" href="https://github.com/hamsal">Hamzeh Alsalhi</a> with thanks
to <a href="#id58"><span class="problematic" id="id59">`Rohit Sivaprasad`_</span></a>), as well as evaluation metrics (by
<a class="reference external" href="http://joelnothman.com">Joel Nothman</a>).</li>
<li>Add <tt class="docutils literal"><span class="pre">sample_weight</span></tt> parameter to <cite>metrics.jaccard_similarity_score</cite>.
By <cite>Jatin Shah</cite>.</li>
<li>Add support for multiclass in <cite>metrics.hinge_loss</cite>. Added <tt class="docutils literal"><span class="pre">labels=None</span></tt>
as optional paramter. By <cite>Saurabh Jha</cite>.</li>
<li>Add <tt class="docutils literal"><span class="pre">sample_weight</span></tt> parameter to <cite>metrics.hinge_loss</cite>.
By <cite>Saurabh Jha</cite>.</li>
<li>Add <tt class="docutils literal"><span class="pre">multi_class="multinomial"</span></tt> option in
<a class="reference internal" href="modules/generated/sklearn.linear_model.LogisticRegression.html#sklearn.linear_model.LogisticRegression" title="sklearn.linear_model.LogisticRegression"><tt class="xref py py-class docutils literal"><span class="pre">linear_model.LogisticRegression</span></tt></a> to implement a Logistic
Regression solver that minimizes the cross-entropy or multinomial loss
instead of the default One-vs-Rest setting. Supports <cite>lbfgs</cite> and
<cite>newton-cg</cite> solvers. By <a class="reference external" href="https://github.com/larsmans">Lars Buitinck</a> and <a class="reference external" href="https://manojbits.wordpress.com">Manoj Kumar</a>. Solver option
<cite>newton-cg</cite> by Simon Wu.</li>
<li><tt class="docutils literal"><span class="pre">DictVectorizer</span></tt> can now perform <tt class="docutils literal"><span class="pre">fit_transform</span></tt> on an iterable in a
single pass, when giving the option <tt class="docutils literal"><span class="pre">sort=False</span></tt>. By Dan Blanchard.</li>
<li><tt class="xref py py-class docutils literal"><span class="pre">GridSearchCV</span></tt> and <tt class="xref py py-class docutils literal"><span class="pre">RandomizedSearchCV</span></tt> can now be
configured to work with estimators that may fail and raise errors on
individual folds. This option is controlled by the <cite>error_score</cite>
parameter. This does not affect errors raised on re-fit. By
<a class="reference external" href="https://github.com/romaniukm">Michal Romaniuk</a>.</li>
<li>Add <tt class="docutils literal"><span class="pre">digits</span></tt> parameter to <cite>metrics.classification_report</cite> to allow
report to show different precision of floating point numbers. By
<a class="reference external" href="https://github.com/agileminor">Ian Gilmore</a>.</li>
<li>Add a quantile prediction strategy to the <a class="reference internal" href="modules/generated/sklearn.dummy.DummyRegressor.html#sklearn.dummy.DummyRegressor" title="sklearn.dummy.DummyRegressor"><tt class="xref py py-class docutils literal"><span class="pre">dummy.DummyRegressor</span></tt></a>.
By <a class="reference external" href="https://github.com/staple">Aaron Staple</a>.</li>
<li>Add <tt class="docutils literal"><span class="pre">handle_unknown</span></tt> option to <a class="reference internal" href="modules/generated/sklearn.preprocessing.OneHotEncoder.html#sklearn.preprocessing.OneHotEncoder" title="sklearn.preprocessing.OneHotEncoder"><tt class="xref py py-class docutils literal"><span class="pre">preprocessing.OneHotEncoder</span></tt></a> to
handle unknown categorical features more gracefully during transform.
By <a class="reference external" href="https://manojbits.wordpress.com">Manoj Kumar</a>.</li>
<li>Added support for sparse input data to decision trees and their ensembles.
By <a class="reference external" href="https://github.com/fareshedyati">Fares Hedyati</a> and <a class="reference external" href="http://www.ajoly.org">Arnaud Joly</a>.</li>
<li>Optimized <a class="reference internal" href="modules/generated/sklearn.cluster.AffinityPropagation.html#sklearn.cluster.AffinityPropagation" title="sklearn.cluster.AffinityPropagation"><tt class="xref py py-class docutils literal"><span class="pre">cluster.AffinityPropagation</span></tt></a> by reducing the number of
memory allocations of large temporary data-structures. By <a class="reference external" href="https://www.ocf.berkeley.edu/~antonyl/">Antony Lee</a>.</li>
<li>Parellization of the computation of feature importances in random forest.
By <a class="reference external" href="http://twitter.com/ogrisel">Olivier Grisel</a> and <a class="reference external" href="http://www.ajoly.org">Arnaud Joly</a>.</li>
<li>Add <tt class="docutils literal"><span class="pre">n_iter_</span></tt> attribute to estimators that accept a <tt class="docutils literal"><span class="pre">max_iter</span></tt> attribute
in their constructor. By <a class="reference external" href="https://manojbits.wordpress.com">Manoj Kumar</a>.</li>
<li><a class="reference internal" href="modules/generated/sklearn.neighbors.kneighbors_graph.html#sklearn.neighbors.kneighbors_graph" title="sklearn.neighbors.kneighbors_graph"><tt class="xref py py-func docutils literal"><span class="pre">neighbors.kneighbors_graph</span></tt></a> and <tt class="xref py py-func docutils literal"><span class="pre">radius_neighbors_graph</span></tt>
support non-Euclidean metrics. By <a class="reference external" href="https://manojbits.wordpress.com">Manoj Kumar</a></li>
<li>Parameter <tt class="docutils literal"><span class="pre">connectivity</span></tt> in <a class="reference internal" href="modules/generated/sklearn.cluster.AgglomerativeClustering.html#sklearn.cluster.AgglomerativeClustering" title="sklearn.cluster.AgglomerativeClustering"><tt class="xref py py-class docutils literal"><span class="pre">cluster.AgglomerativeClustering</span></tt></a>
and family now accept callables that return a connectivity matrix.
By <a class="reference external" href="https://manojbits.wordpress.com">Manoj Kumar</a>.</li>
<li>Sparse support for <tt class="xref py py-func docutils literal"><span class="pre">paired_distances</span></tt>. By <a class="reference external" href="http://joelnothman.com">Joel Nothman</a>.</li>
<li>DBSCAN now supports sparse input and sample weights, and should be
faster in general. By <a class="reference external" href="http://joelnothman.com">Joel Nothman</a>.</li>
<li>Add <tt class="docutils literal"><span class="pre">class_weight</span></tt> parameter to automatically weight samples by class
frequency for <a class="reference internal" href="modules/generated/sklearn.ensemble.RandomForestClassifier.html#sklearn.ensemble.RandomForestClassifier" title="sklearn.ensemble.RandomForestClassifier"><tt class="xref py py-class docutils literal"><span class="pre">ensemble.RandomForestClassifier</span></tt></a>,
<a class="reference internal" href="modules/generated/sklearn.tree.DecisionTreeClassifier.html#sklearn.tree.DecisionTreeClassifier" title="sklearn.tree.DecisionTreeClassifier"><tt class="xref py py-class docutils literal"><span class="pre">tree.DecisionTreeClassifier</span></tt></a>, <a class="reference internal" href="modules/generated/sklearn.ensemble.ExtraTreesClassifier.html#sklearn.ensemble.ExtraTreesClassifier" title="sklearn.ensemble.ExtraTreesClassifier"><tt class="xref py py-class docutils literal"><span class="pre">ensemble.ExtraTreesClassifier</span></tt></a>
and <a class="reference internal" href="modules/generated/sklearn.tree.ExtraTreeClassifier.html#sklearn.tree.ExtraTreeClassifier" title="sklearn.tree.ExtraTreeClassifier"><tt class="xref py py-class docutils literal"><span class="pre">tree.ExtraTreeClassifier</span></tt></a>. By <a class="reference external" href="http://trevorstephens.com/">Trevor Stephens</a>.</li>
<li><a class="reference internal" href="modules/generated/sklearn.grid_search.RandomizedSearchCV.html#sklearn.grid_search.RandomizedSearchCV" title="sklearn.grid_search.RandomizedSearchCV"><tt class="xref py py-class docutils literal"><span class="pre">grid_search.RandomizedSearchCV</span></tt></a> now does sampling without
replacement if all parameters are given as lists. By <a class="reference external" href="http://peekaboo-vision.blogspot.com">Andreas Müller</a>.</li>
<li>Parallelized calculation of <tt class="xref py py-func docutils literal"><span class="pre">pairwise_distances</span></tt> is now supported
for scipy metrics and custom callables. By <a class="reference external" href="http://joelnothman.com">Joel Nothman</a>.</li>
<li>Allow the fitting and scoring of all clustering algorithms in
<a class="reference internal" href="modules/generated/sklearn.pipeline.Pipeline.html#sklearn.pipeline.Pipeline" title="sklearn.pipeline.Pipeline"><tt class="xref py py-class docutils literal"><span class="pre">pipeline.Pipeline</span></tt></a>. By <a class="reference external" href="http://peekaboo-vision.blogspot.com">Andreas Müller</a>.</li>
<li>More robust seeding and improved error messages in <a class="reference internal" href="modules/generated/sklearn.cluster.MeanShift.html#sklearn.cluster.MeanShift" title="sklearn.cluster.MeanShift"><tt class="xref py py-class docutils literal"><span class="pre">cluster.MeanShift</span></tt></a>
by <a class="reference external" href="http://peekaboo-vision.blogspot.com">Andreas Müller</a>.</li>
<li>Make the <tt class="xref py py-class docutils literal"><span class="pre">GMM</span></tt> stopping criterion less dependent on the number of
samples by thresholding the average log-likelihood change instead of its
sum over all samples. By <a href="#id60"><span class="problematic" id="id61">`Hervé Bredin`_</span></a></li>
</ul>
</div></blockquote>
</div>
<div class="section" id="documentation-improvements">
<h3>Documentation improvements<a class="headerlink" href="#documentation-improvements" title="Permalink to this headline">¶</a></h3>
<blockquote>
<div><ul class="simple">
<li>Added example of using <tt class="xref py py-class docutils literal"><span class="pre">FeatureUnion</span></tt> for heterogeneous input.
By <a class="reference external" href="https://github.com/mrterry">Matt Terry</a></li>
<li>Documentation on scorers was improved, to highlight the handling of loss
functions. By <a class="reference external" href="https://github.com/MattpSoftware">Matt Pico</a>.</li>
<li>A discrepancy between liblinear output and scikit-learn’s wrappers
is now noted. By <a class="reference external" href="https://manojbits.wordpress.com">Manoj Kumar</a>.</li>
<li>Improved documentation generation: examples referring to a class or
function are now shown in a gallery on the class/function’s API reference
page. By <a class="reference external" href="http://joelnothman.com">Joel Nothman</a>.</li>
<li>More explicit documentation of sample generators and of data
transformation. By <a class="reference external" href="http://joelnothman.com">Joel Nothman</a>.</li>
<li><a class="reference internal" href="modules/generated/sklearn.neighbors.BallTree.html#sklearn.neighbors.BallTree" title="sklearn.neighbors.BallTree"><tt class="xref py py-class docutils literal"><span class="pre">sklearn.neighbors.BallTree</span></tt></a> and <a class="reference internal" href="modules/generated/sklearn.neighbors.KDTree.html#sklearn.neighbors.KDTree" title="sklearn.neighbors.KDTree"><tt class="xref py py-class docutils literal"><span class="pre">sklearn.neighbors.KDTree</span></tt></a>
used to point to empty pages stating that they are aliases of BinaryTree.
This has been fixed to show the correct class docs. By <a class="reference external" href="https://manojbits.wordpress.com">Manoj Kumar</a>.</li>
<li>Added silhouette plots for analysis of KMeans clustering using
<a class="reference internal" href="modules/generated/sklearn.metrics.silhouette_samples.html#sklearn.metrics.silhouette_samples" title="sklearn.metrics.silhouette_samples"><tt class="xref py py-func docutils literal"><span class="pre">metrics.silhouette_samples</span></tt></a> and <a class="reference internal" href="modules/generated/sklearn.metrics.silhouette_score.html#sklearn.metrics.silhouette_score" title="sklearn.metrics.silhouette_score"><tt class="xref py py-func docutils literal"><span class="pre">metrics.silhouette_score</span></tt></a>.
See <em class="xref std std-ref">examples_cluster_plot_kmeans_silhouette_analysis.py</em></li>
</ul>
</div></blockquote>
</div>
<div class="section" id="bug-fixes">
<h3>Bug fixes<a class="headerlink" href="#bug-fixes" title="Permalink to this headline">¶</a></h3>
<blockquote>
<div><ul class="simple">
<li>Metaestimators now support ducktyping for the presence of <tt class="docutils literal"><span class="pre">decision_function</span></tt>,
<tt class="docutils literal"><span class="pre">predict_proba</span></tt> and other methods. This fixes behavior of
<a class="reference internal" href="modules/generated/sklearn.grid_search.GridSearchCV.html#sklearn.grid_search.GridSearchCV" title="sklearn.grid_search.GridSearchCV"><tt class="xref py py-class docutils literal"><span class="pre">grid_search.GridSearchCV</span></tt></a>,
<a class="reference internal" href="modules/generated/sklearn.grid_search.RandomizedSearchCV.html#sklearn.grid_search.RandomizedSearchCV" title="sklearn.grid_search.RandomizedSearchCV"><tt class="xref py py-class docutils literal"><span class="pre">grid_search.RandomizedSearchCV</span></tt></a>, <a class="reference internal" href="modules/generated/sklearn.pipeline.Pipeline.html#sklearn.pipeline.Pipeline" title="sklearn.pipeline.Pipeline"><tt class="xref py py-class docutils literal"><span class="pre">pipeline.Pipeline</span></tt></a>,
<a class="reference internal" href="modules/generated/sklearn.feature_selection.RFE.html#sklearn.feature_selection.RFE" title="sklearn.feature_selection.RFE"><tt class="xref py py-class docutils literal"><span class="pre">feature_selection.RFE</span></tt></a>, <a class="reference internal" href="modules/generated/sklearn.feature_selection.RFECV.html#sklearn.feature_selection.RFECV" title="sklearn.feature_selection.RFECV"><tt class="xref py py-class docutils literal"><span class="pre">feature_selection.RFECV</span></tt></a> when nested.
By <a class="reference external" href="http://joelnothman.com">Joel Nothman</a></li>
<li>The <tt class="docutils literal"><span class="pre">scoring</span></tt> attribute of grid-search and cross-validation methods is no longer</li>
</ul>
<blockquote>
<div>ignored when a <a class="reference internal" href="modules/generated/sklearn.grid_search.GridSearchCV.html#sklearn.grid_search.GridSearchCV" title="sklearn.grid_search.GridSearchCV"><tt class="xref py py-class docutils literal"><span class="pre">grid_search.GridSearchCV</span></tt></a> is given as a base estimator or
the base estimator doesn’t have predict.</div></blockquote>
<ul class="simple">
<li>The function <tt class="xref py py-func docutils literal"><span class="pre">hierarchical.ward_tree</span></tt> now returns the children in
the same order for both the structured and unstructured versions. By
<a class="reference external" href="http://www.mvdoc.me">Matteo Visconti di Oleggio Castello</a>.</li>
<li><a class="reference internal" href="modules/generated/sklearn.feature_selection.RFECV.html#sklearn.feature_selection.RFECV" title="sklearn.feature_selection.RFECV"><tt class="xref py py-class docutils literal"><span class="pre">feature_selection.RFECV</span></tt></a> now correctly handles cases when
<tt class="docutils literal"><span class="pre">step</span></tt> is not equal to 1. By <a class="reference external" href="https://github.com/nmayorov">Nikolay Mayorov</a></li>
<li>The <a class="reference internal" href="modules/generated/sklearn.decomposition.PCA.html#sklearn.decomposition.PCA" title="sklearn.decomposition.PCA"><tt class="xref py py-class docutils literal"><span class="pre">decomposition.PCA</span></tt></a> now undoes whitening in its</li>
</ul>
<blockquote>
<div><tt class="docutils literal"><span class="pre">inverse_transform</span></tt>. Also, its <tt class="docutils literal"><span class="pre">components_</span></tt> now always have unit
length. By Michael Eickenberg.</div></blockquote>
<ul class="simple">
<li>Fix incomplete download of the dataset when
<tt class="xref py py-func docutils literal"><span class="pre">datasets.download_20newsgroups</span></tt> is called. By <a class="reference external" href="https://manojbits.wordpress.com">Manoj Kumar</a>.</li>
<li>Various fixes to the Gaussian processes subpackage by Vincent Dubourg
and Jan Hendrik Metzen.</li>
<li>Calling <tt class="docutils literal"><span class="pre">partial_fit</span></tt> with <tt class="docutils literal"><span class="pre">class_weight=='auto'</span></tt> throws an
appropriate error message and suggests a work around.
By <a class="reference external" href="https://github.com/dsullivan7">Danny Sullivan</a>.</li>
<li><a class="reference internal" href="modules/generated/sklearn.kernel_approximation.RBFSampler.html#sklearn.kernel_approximation.RBFSampler" title="sklearn.kernel_approximation.RBFSampler"><tt class="xref py py-class docutils literal"><span class="pre">RBFSampler</span></tt></a> with <tt class="docutils literal"><span class="pre">gamma=g</span></tt>
formerly approximated <a class="reference internal" href="modules/generated/sklearn.metrics.pairwise.rbf_kernel.html#sklearn.metrics.pairwise.rbf_kernel" title="sklearn.metrics.pairwise.rbf_kernel"><tt class="xref py py-func docutils literal"><span class="pre">rbf_kernel</span></tt></a>
with <tt class="docutils literal"><span class="pre">gamma=g/2.</span></tt>; the definition of <tt class="docutils literal"><span class="pre">gamma</span></tt> is now consistent,
which may substantially change your results if you use a fixed value.
(If you cross-validated over <tt class="docutils literal"><span class="pre">gamma</span></tt>, it probably doesn’t matter
too much.) By <a class="reference external" href="https://github.com/dougalsutherland">Dougal Sutherland</a>.</li>
<li>Pipeline object delegate the <tt class="docutils literal"><span class="pre">classes_</span></tt> attribute to the underlying
estimator. It allows for instance to make bagging of a pipeline object.
By <a class="reference external" href="http://www.ajoly.org">Arnaud Joly</a></li>
<li><a class="reference internal" href="modules/generated/sklearn.neighbors.NearestCentroid.html#sklearn.neighbors.NearestCentroid" title="sklearn.neighbors.NearestCentroid"><tt class="xref py py-class docutils literal"><span class="pre">neighbors.NearestCentroid</span></tt></a> now uses the median as the centroid
when metric is set to <tt class="docutils literal"><span class="pre">manhattan</span></tt>. It was using the mean before.
By <a class="reference external" href="https://manojbits.wordpress.com">Manoj Kumar</a></li>
<li>Fix numerical stability issues in <a class="reference internal" href="modules/generated/sklearn.linear_model.SGDClassifier.html#sklearn.linear_model.SGDClassifier" title="sklearn.linear_model.SGDClassifier"><tt class="xref py py-class docutils literal"><span class="pre">linear_model.SGDClassifier</span></tt></a>
and <a class="reference internal" href="modules/generated/sklearn.linear_model.SGDRegressor.html#sklearn.linear_model.SGDRegressor" title="sklearn.linear_model.SGDRegressor"><tt class="xref py py-class docutils literal"><span class="pre">linear_model.SGDRegressor</span></tt></a> by clipping large gradients and
ensuring that weight decay rescaling is always positive (for large
l2 regularization and large learning rate values).
By <a class="reference external" href="http://twitter.com/ogrisel">Olivier Grisel</a></li>
<li>When <cite>compute_full_tree</cite> is set to “auto”, the full tree is
built when n_clusters is high and is early stopped when n_clusters is
low, while the behavior should be vice-versa in
<a class="reference internal" href="modules/generated/sklearn.cluster.AgglomerativeClustering.html#sklearn.cluster.AgglomerativeClustering" title="sklearn.cluster.AgglomerativeClustering"><tt class="xref py py-class docutils literal"><span class="pre">cluster.AgglomerativeClustering</span></tt></a> (and friends).
This has been fixed By <a class="reference external" href="https://manojbits.wordpress.com">Manoj Kumar</a></li>
<li>Fix lazy centering of data in <tt class="xref py py-func docutils literal"><span class="pre">linear_model.enet_path</span></tt> and
<a class="reference internal" href="modules/generated/sklearn.linear_model.lasso_path.html#sklearn.linear_model.lasso_path" title="sklearn.linear_model.lasso_path"><tt class="xref py py-func docutils literal"><span class="pre">linear_model.lasso_path</span></tt></a>. It was centered around one. It has
been changed to be centered around the origin. By <a class="reference external" href="https://manojbits.wordpress.com">Manoj Kumar</a></li>
<li>Fix handling of precomputed affinity matrices in
<a class="reference internal" href="modules/generated/sklearn.cluster.AgglomerativeClustering.html#sklearn.cluster.AgglomerativeClustering" title="sklearn.cluster.AgglomerativeClustering"><tt class="xref py py-class docutils literal"><span class="pre">cluster.AgglomerativeClustering</span></tt></a> when using connectivity
constraints. By <a class="reference external" href="https://github.com/cathydeng">Cathy Deng</a></li>
<li>Correct <tt class="docutils literal"><span class="pre">partial_fit</span></tt> handling of <tt class="docutils literal"><span class="pre">class_prior</span></tt> for
<a class="reference internal" href="modules/generated/sklearn.naive_bayes.MultinomialNB.html#sklearn.naive_bayes.MultinomialNB" title="sklearn.naive_bayes.MultinomialNB"><tt class="xref py py-class docutils literal"><span class="pre">sklearn.naive_bayes.MultinomialNB</span></tt></a> and
<a class="reference internal" href="modules/generated/sklearn.naive_bayes.BernoulliNB.html#sklearn.naive_bayes.BernoulliNB" title="sklearn.naive_bayes.BernoulliNB"><tt class="xref py py-class docutils literal"><span class="pre">sklearn.naive_bayes.BernoulliNB</span></tt></a>. By <a class="reference external" href="http://trevorstephens.com/">Trevor Stephens</a>.</li>
<li>Fixed a crash in <a class="reference internal" href="modules/generated/sklearn.metrics.precision_recall_fscore_support.html#sklearn.metrics.precision_recall_fscore_support" title="sklearn.metrics.precision_recall_fscore_support"><tt class="xref py py-func docutils literal"><span class="pre">metrics.precision_recall_fscore_support</span></tt></a>
when using unsorted <tt class="docutils literal"><span class="pre">labels</span></tt> in the multi-label setting.
By <a class="reference external" href="http://peekaboo-vision.blogspot.com">Andreas Müller</a>.</li>
<li>Avoid skipping the first nearest neighbor in the methods <tt class="docutils literal"><span class="pre">radius_neighbors</span></tt>,
<tt class="docutils literal"><span class="pre">kneighbors</span></tt>, <tt class="docutils literal"><span class="pre">kneighbors_graph</span></tt> and <tt class="docutils literal"><span class="pre">radius_neighbors_graph</span></tt> in
<a class="reference internal" href="modules/generated/sklearn.neighbors.NearestNeighbors.html#sklearn.neighbors.NearestNeighbors" title="sklearn.neighbors.NearestNeighbors"><tt class="xref py py-class docutils literal"><span class="pre">sklearn.neighbors.NearestNeighbors</span></tt></a> and family, when the query
data is not the same as fit data. By <a class="reference external" href="https://manojbits.wordpress.com">Manoj Kumar</a>.</li>
<li>Fix log-density calculation in the <a class="reference internal" href="modules/generated/sklearn.mixture.GMM.html#sklearn.mixture.GMM" title="sklearn.mixture.GMM"><tt class="xref py py-class docutils literal"><span class="pre">mixture.GMM</span></tt></a> with
tied covariance. By <a class="reference external" href="http://dawsonresearch.com">Will Dawson</a></li>
</ul>
</div></blockquote>
</div>
</div>
<div class="section" id="api-changes-summary">
<h2>API changes summary<a class="headerlink" href="#api-changes-summary" title="Permalink to this headline">¶</a></h2>
<blockquote>
<div><ul>
<li><p class="first"><a class="reference internal" href="modules/generated/sklearn.grid_search.GridSearchCV.html#sklearn.grid_search.GridSearchCV" title="sklearn.grid_search.GridSearchCV"><tt class="xref py py-class docutils literal"><span class="pre">GridSearchCV</span></tt></a> and
<a class="reference internal" href="modules/generated/sklearn.cross_validation.cross_val_score.html#sklearn.cross_validation.cross_val_score" title="sklearn.cross_validation.cross_val_score"><tt class="xref py py-func docutils literal"><span class="pre">cross_val_score</span></tt></a> and other
meta-estimators don’t convert pandas DataFrames into arrays any more,
allowing DataFrame specific operations in custom estimators.</p>
</li>
<li><p class="first"><tt class="xref py py-func docutils literal"><span class="pre">multiclass.fit_ovr</span></tt>, <tt class="xref py py-func docutils literal"><span class="pre">multiclass.predict_ovr</span></tt>,
<tt class="xref py py-func docutils literal"><span class="pre">predict_proba_ovr</span></tt>,
<tt class="xref py py-func docutils literal"><span class="pre">multiclass.fit_ovo</span></tt>, <tt class="xref py py-func docutils literal"><span class="pre">multiclass.predict_ovo</span></tt>,
<tt class="xref py py-func docutils literal"><span class="pre">multiclass.fit_ecoc</span></tt> and <tt class="xref py py-func docutils literal"><span class="pre">multiclass.predict_ecoc</span></tt>
are deprecated. Use the underlying estimators instead.</p>
</li>
<li><p class="first">Nearest neighbors estimators used to take arbitrary keyword arguments
and pass these to their distance metric. This will no longer be supported
in scikit-learn 0.18; use the <tt class="docutils literal"><span class="pre">metric_params</span></tt> argument instead.</p>
</li>
<li><dl class="first docutils">
<dt><cite>n_jobs</cite> parameter of the fit method shifted to the constructor of the</dt>
<dd><p class="first last">LinearRegression class.</p>
</dd>
</dl>
</li>
<li><p class="first">The <tt class="docutils literal"><span class="pre">predict_proba</span></tt> method of <a class="reference internal" href="modules/generated/sklearn.multiclass.OneVsRestClassifier.html#sklearn.multiclass.OneVsRestClassifier" title="sklearn.multiclass.OneVsRestClassifier"><tt class="xref py py-class docutils literal"><span class="pre">multiclass.OneVsRestClassifier</span></tt></a>
now returns two probabilities per sample in the multiclass case; this
is consistent with other estimators and with the method’s documentation,
but previous versions accidentally returned only the positive
probability. Fixed by Will Lamond and <a class="reference external" href="https://github.com/larsmans">Lars Buitinck</a>.</p>
</li>
<li><p class="first">Change default value of precompute in <tt class="xref py py-class docutils literal"><span class="pre">ElasticNet</span></tt> and <tt class="xref py py-class docutils literal"><span class="pre">Lasso</span></tt>
to False. Setting precompute to “auto” was found to be slower when
n_samples > n_features since the computation of the Gram matrix is
computationally expensive and outweighs the benefit of fitting the Gram
for just one alpha.
<tt class="docutils literal"><span class="pre">precompute="auto"</span></tt> is now deprecated and will be removed in 0.18
By <a class="reference external" href="https://manojbits.wordpress.com">Manoj Kumar</a>.</p>
</li>
<li><p class="first">Expose <tt class="docutils literal"><span class="pre">positive</span></tt> option in <tt class="xref py py-func docutils literal"><span class="pre">linear_model.enet_path</span></tt> and
<tt class="xref py py-func docutils literal"><span class="pre">linear_model.enet_path</span></tt> which constrains coefficients to be
positive. By <a class="reference external" href="https://manojbits.wordpress.com">Manoj Kumar</a>.</p>
</li>
<li><p class="first">Users should now supply an explicit <tt class="docutils literal"><span class="pre">average</span></tt> parameter to
<a class="reference internal" href="modules/generated/sklearn.metrics.f1_score.html#sklearn.metrics.f1_score" title="sklearn.metrics.f1_score"><tt class="xref py py-func docutils literal"><span class="pre">sklearn.metrics.f1_score</span></tt></a>, <a class="reference internal" href="modules/generated/sklearn.metrics.fbeta_score.html#sklearn.metrics.fbeta_score" title="sklearn.metrics.fbeta_score"><tt class="xref py py-func docutils literal"><span class="pre">sklearn.metrics.fbeta_score</span></tt></a>,
<a class="reference internal" href="modules/generated/sklearn.metrics.recall_score.html#sklearn.metrics.recall_score" title="sklearn.metrics.recall_score"><tt class="xref py py-func docutils literal"><span class="pre">sklearn.metrics.recall_score</span></tt></a> and
<a class="reference internal" href="modules/generated/sklearn.metrics.precision_score.html#sklearn.metrics.precision_score" title="sklearn.metrics.precision_score"><tt class="xref py py-func docutils literal"><span class="pre">sklearn.metrics.precision_score</span></tt></a> when performing multiclass
or multilabel (i.e. not binary) classification. By <a class="reference external" href="http://joelnothman.com">Joel Nothman</a>.</p>
</li>
<li><p class="first"><cite>scoring</cite> parameter for cross validation now accepts <cite>‘f1_micro’</cite>,
<cite>‘f1_macro’</cite> or <cite>‘f1_weighted’</cite>. <cite>‘f1’</cite> is now for binary classification
only. Similar changes apply to <cite>‘precision’</cite> and <cite>‘recall’</cite>.
By <a class="reference external" href="http://joelnothman.com">Joel Nothman</a>.</p>
</li>
<li><p class="first">The <tt class="docutils literal"><span class="pre">fit_intercept</span></tt>, <tt class="docutils literal"><span class="pre">normalize</span></tt> and <tt class="docutils literal"><span class="pre">return_models</span></tt> parameters in
<tt class="xref py py-func docutils literal"><span class="pre">linear_model.enet_path</span></tt> and <a class="reference internal" href="modules/generated/sklearn.linear_model.lasso_path.html#sklearn.linear_model.lasso_path" title="sklearn.linear_model.lasso_path"><tt class="xref py py-func docutils literal"><span class="pre">linear_model.lasso_path</span></tt></a> have
been removed. They were deprecated since 0.14</p>
</li>
<li><p class="first">From now onwards, all estimators will uniformly raise <tt class="docutils literal"><span class="pre">NotFittedError</span></tt>
(<tt class="xref py py-class docutils literal"><span class="pre">utils.validation.NotFittedError</span></tt>), when any of the <tt class="docutils literal"><span class="pre">predict</span></tt>
like methods are called before the model is fit. By <a class="reference external" href="https://github.com/ragv">Raghav R V</a>.</p>
</li>
<li><p class="first">Input data validation was refactored for more consistent input
validation. The <tt class="docutils literal"><span class="pre">check_arrays</span></tt> function was replaced by <tt class="docutils literal"><span class="pre">check_array</span></tt>
and <tt class="docutils literal"><span class="pre">check_X_y</span></tt>. By <a class="reference external" href="http://peekaboo-vision.blogspot.com">Andreas Müller</a>.</p>
</li>
<li><p class="first">Allow <tt class="docutils literal"><span class="pre">X=None</span></tt> in the methods <tt class="docutils literal"><span class="pre">radius_neighbors</span></tt>, <tt class="docutils literal"><span class="pre">kneighbors</span></tt>,
<tt class="docutils literal"><span class="pre">kneighbors_graph</span></tt> and <tt class="docutils literal"><span class="pre">radius_neighbors_graph</span></tt> in
<a class="reference internal" href="modules/generated/sklearn.neighbors.NearestNeighbors.html#sklearn.neighbors.NearestNeighbors" title="sklearn.neighbors.NearestNeighbors"><tt class="xref py py-class docutils literal"><span class="pre">sklearn.neighbors.NearestNeighbors</span></tt></a> and family. If set to None,
then for every sample this avoids setting the sample itself as the
first nearest neighbor. By <a class="reference external" href="https://manojbits.wordpress.com">Manoj Kumar</a>.</p>
</li>
<li><p class="first">Add parameter <tt class="docutils literal"><span class="pre">include_self</span></tt> in <a class="reference internal" href="modules/generated/sklearn.neighbors.kneighbors_graph.html#sklearn.neighbors.kneighbors_graph" title="sklearn.neighbors.kneighbors_graph"><tt class="xref py py-func docutils literal"><span class="pre">neighbors.kneighbors_graph</span></tt></a>
and <a class="reference internal" href="modules/generated/sklearn.neighbors.radius_neighbors_graph.html#sklearn.neighbors.radius_neighbors_graph" title="sklearn.neighbors.radius_neighbors_graph"><tt class="xref py py-func docutils literal"><span class="pre">neighbors.radius_neighbors_graph</span></tt></a> which has to be explicitly
set by the user. If set to True, then the sample itself is considered
as the first nearest neighbor.</p>
</li>
<li><p class="first"><cite>thresh</cite> parameter is deprecated in favor of new <cite>tol</cite> parameter in
<tt class="xref py py-class docutils literal"><span class="pre">GMM</span></tt>. See <cite>Enhancements</cite> section for details. By <a href="#id62"><span class="problematic" id="id63">`Hervé Bredin`_</span></a>.</p>
</li>
</ul>
</div></blockquote>
</div>
</div>
<div class="section" id="changes-0-15-2">
<span id="id2"></span><h1>0.15.2<a class="headerlink" href="#changes-0-15-2" title="Permalink to this headline">¶</a></h1>
<div class="section" id="id3">
<h2>Bug fixes<a class="headerlink" href="#id3" title="Permalink to this headline">¶</a></h2>
<blockquote>
<div><ul class="simple">
<li>Fixed handling of the <tt class="docutils literal"><span class="pre">p</span></tt> parameter of the Minkowski distance that was
previously ignored in nearest neighbors models. By <a class="reference external" href="https://github.com/nmayorov">Nikolay Mayorov</a>.</li>
<li>Fixed duplicated alphas in <a class="reference internal" href="modules/generated/sklearn.linear_model.LassoLars.html#sklearn.linear_model.LassoLars" title="sklearn.linear_model.LassoLars"><tt class="xref py py-class docutils literal"><span class="pre">linear_model.LassoLars</span></tt></a> with early
stopping on 32 bit Python. By <a class="reference external" href="http://twitter.com/ogrisel">Olivier Grisel</a> and <a class="reference external" href="http://fseoane.net/blog/">Fabian Pedregosa</a>.</li>
<li>Fixed the build under Windows when scikit-learn is built with MSVC while
NumPy is built with MinGW. By <a class="reference external" href="http://twitter.com/ogrisel">Olivier Grisel</a> and Federico Vaggi.</li>
<li>Fixed an array index overflow bug in the coordinate descent solver. By
<a class="reference external" href="http://gael-varoquaux.info">Gael Varoquaux</a>.</li>
<li>Better handling of numpy 1.9 deprecation warnings. By <a class="reference external" href="http://gael-varoquaux.info">Gael Varoquaux</a>.</li>
<li>Removed unnecessary data copy in <a class="reference internal" href="modules/generated/sklearn.cluster.KMeans.html#sklearn.cluster.KMeans" title="sklearn.cluster.KMeans"><tt class="xref py py-class docutils literal"><span class="pre">cluster.KMeans</span></tt></a>.
By <a class="reference external" href="http://gael-varoquaux.info">Gael Varoquaux</a>.</li>
<li>Explicitly close open files to avoid <tt class="docutils literal"><span class="pre">ResourceWarnings</span></tt> under Python 3.
By Calvin Giles.</li>
<li>The <tt class="docutils literal"><span class="pre">transform</span></tt> of <a class="reference internal" href="modules/generated/sklearn.lda.LDA.html#sklearn.lda.LDA" title="sklearn.lda.LDA"><tt class="xref py py-class docutils literal"><span class="pre">lda.LDA</span></tt></a> now projects the input on the most
discriminant directions. By Martin Billinger.</li>
<li>Fixed potential overflow in <tt class="docutils literal"><span class="pre">_tree.safe_realloc</span></tt> by <a class="reference external" href="https://github.com/larsmans">Lars Buitinck</a>.</li>
<li>Performance optimization in <a class="reference internal" href="modules/generated/sklearn.isotonic.IsotonicRegression.html#sklearn.isotonic.IsotonicRegression" title="sklearn.isotonic.IsotonicRegression"><tt class="xref py py-class docutils literal"><span class="pre">isotonic.IsotonicRegression</span></tt></a>.
By Robert Bradshaw.</li>
<li><tt class="docutils literal"><span class="pre">nose</span></tt> is non-longer a runtime dependency to import <tt class="docutils literal"><span class="pre">sklearn</span></tt>, only for
running the tests. By <a class="reference external" href="http://joelnothman.com">Joel Nothman</a>.</li>
<li>Many documentation and website fixes by <a class="reference external" href="http://joelnothman.com">Joel Nothman</a>, <a class="reference external" href="https://github.com/larsmans">Lars Buitinck</a>
<a class="reference external" href="https://github.com/MattpSoftware">Matt Pico</a>, and others.</li>
</ul>
</div></blockquote>
</div>
</div>
<div class="section" id="changes-0-15-1">
<span id="id4"></span><h1>0.15.1<a class="headerlink" href="#changes-0-15-1" title="Permalink to this headline">¶</a></h1>
<div class="section" id="id5">
<h2>Bug fixes<a class="headerlink" href="#id5" title="Permalink to this headline">¶</a></h2>
<blockquote>
<div><ul class="simple">
<li>Made <a class="reference internal" href="modules/generated/sklearn.cross_validation.cross_val_score.html#sklearn.cross_validation.cross_val_score" title="sklearn.cross_validation.cross_val_score"><tt class="xref py py-func docutils literal"><span class="pre">cross_validation.cross_val_score</span></tt></a> use
<a class="reference internal" href="modules/generated/sklearn.cross_validation.KFold.html#sklearn.cross_validation.KFold" title="sklearn.cross_validation.KFold"><tt class="xref py py-class docutils literal"><span class="pre">cross_validation.KFold</span></tt></a> instead of
<a class="reference internal" href="modules/generated/sklearn.cross_validation.StratifiedKFold.html#sklearn.cross_validation.StratifiedKFold" title="sklearn.cross_validation.StratifiedKFold"><tt class="xref py py-class docutils literal"><span class="pre">cross_validation.StratifiedKFold</span></tt></a> on multi-output classification
problems. By <a class="reference external" href="https://github.com/nmayorov">Nikolay Mayorov</a>.</li>
<li>Support unseen labels <a class="reference internal" href="modules/generated/sklearn.preprocessing.LabelBinarizer.html#sklearn.preprocessing.LabelBinarizer" title="sklearn.preprocessing.LabelBinarizer"><tt class="xref py py-class docutils literal"><span class="pre">preprocessing.LabelBinarizer</span></tt></a> to restore
the default behavior of 0.14.1 for backward compatibility. By
<a class="reference external" href="https://github.com/hamsal">Hamzeh Alsalhi</a>.</li>
<li>Fixed the <a class="reference internal" href="modules/generated/sklearn.cluster.KMeans.html#sklearn.cluster.KMeans" title="sklearn.cluster.KMeans"><tt class="xref py py-class docutils literal"><span class="pre">cluster.KMeans</span></tt></a> stopping criterion that prevented early
convergence detection. By Edward Raff and <a class="reference external" href="http://gael-varoquaux.info">Gael Varoquaux</a>.</li>
<li>Fixed the behavior of <a class="reference internal" href="modules/generated/sklearn.multiclass.OneVsOneClassifier.html#sklearn.multiclass.OneVsOneClassifier" title="sklearn.multiclass.OneVsOneClassifier"><tt class="xref py py-class docutils literal"><span class="pre">multiclass.OneVsOneClassifier</span></tt></a>.
in case of ties at the per-class vote level by computing the correct
per-class sum of prediction scores. By <a class="reference external" href="http://peekaboo-vision.blogspot.com">Andreas Müller</a>.</li>
<li>Made <a class="reference internal" href="modules/generated/sklearn.cross_validation.cross_val_score.html#sklearn.cross_validation.cross_val_score" title="sklearn.cross_validation.cross_val_score"><tt class="xref py py-func docutils literal"><span class="pre">cross_validation.cross_val_score</span></tt></a> and
<a class="reference internal" href="modules/generated/sklearn.grid_search.GridSearchCV.html#sklearn.grid_search.GridSearchCV" title="sklearn.grid_search.GridSearchCV"><tt class="xref py py-class docutils literal"><span class="pre">grid_search.GridSearchCV</span></tt></a> accept Python lists as input data.
This is especially useful for cross-validation and model selection of
text processing pipelines. By <a class="reference external" href="http://peekaboo-vision.blogspot.com">Andreas Müller</a>.</li>
<li>Fixed data input checks of most estimators to accept input data that
implements the NumPy <tt class="docutils literal"><span class="pre">__array__</span></tt> protocol. This is the case for
for <tt class="docutils literal"><span class="pre">pandas.Series</span></tt> and <tt class="docutils literal"><span class="pre">pandas.DataFrame</span></tt> in recent versions of
pandas. By <a class="reference external" href="http://gael-varoquaux.info">Gael Varoquaux</a>.</li>
<li>Fixed a regression for <a class="reference internal" href="modules/generated/sklearn.linear_model.SGDClassifier.html#sklearn.linear_model.SGDClassifier" title="sklearn.linear_model.SGDClassifier"><tt class="xref py py-class docutils literal"><span class="pre">linear_model.SGDClassifier</span></tt></a> with
<tt class="docutils literal"><span class="pre">class_weight="auto"</span></tt> on data with non-contiguous labels. By
<a class="reference external" href="http://twitter.com/ogrisel">Olivier Grisel</a>.</li>
</ul>
</div></blockquote>
</div>
</div>
<div class="section" id="changes-0-15">
<span id="id6"></span><h1>0.15<a class="headerlink" href="#changes-0-15" title="Permalink to this headline">¶</a></h1>
<div class="section" id="highlights">
<h2>Highlights<a class="headerlink" href="#highlights" title="Permalink to this headline">¶</a></h2>
<blockquote>
<div><ul class="simple">
<li>Many speed and memory improvements all across the code</li>
<li>Huge speed and memory improvements to random forests (and extra
trees) that also benefit better from parallel computing.</li>
<li>Incremental fit to <a class="reference internal" href="modules/generated/sklearn.neural_network.BernoulliRBM.html#sklearn.neural_network.BernoulliRBM" title="sklearn.neural_network.BernoulliRBM"><tt class="xref py py-class docutils literal"><span class="pre">BernoulliRBM</span></tt></a></li>
<li>Added <a class="reference internal" href="modules/generated/sklearn.cluster.AgglomerativeClustering.html#sklearn.cluster.AgglomerativeClustering" title="sklearn.cluster.AgglomerativeClustering"><tt class="xref py py-class docutils literal"><span class="pre">cluster.AgglomerativeClustering</span></tt></a> for hierarchical
agglomerative clustering with average linkage, complete linkage and
ward strategies.</li>
<li>Added <a class="reference internal" href="modules/generated/sklearn.linear_model.RANSACRegressor.html#sklearn.linear_model.RANSACRegressor" title="sklearn.linear_model.RANSACRegressor"><tt class="xref py py-class docutils literal"><span class="pre">linear_model.RANSACRegressor</span></tt></a> for robust regression
models.</li>
<li>Added dimensionality reduction with <a class="reference internal" href="modules/generated/sklearn.manifold.TSNE.html#sklearn.manifold.TSNE" title="sklearn.manifold.TSNE"><tt class="xref py py-class docutils literal"><span class="pre">manifold.TSNE</span></tt></a> which can be
used to visualize high-dimensional data.</li>
</ul>
</div></blockquote>
</div>
<div class="section" id="id7">
<h2>Changelog<a class="headerlink" href="#id7" title="Permalink to this headline">¶</a></h2>
<div class="section" id="id8">
<h3>New features<a class="headerlink" href="#id8" title="Permalink to this headline">¶</a></h3>
<blockquote>
<div><ul class="simple">
<li>Added <a class="reference internal" href="modules/generated/sklearn.ensemble.BaggingClassifier.html#sklearn.ensemble.BaggingClassifier" title="sklearn.ensemble.BaggingClassifier"><tt class="xref py py-class docutils literal"><span class="pre">ensemble.BaggingClassifier</span></tt></a> and
<a class="reference internal" href="modules/generated/sklearn.ensemble.BaggingRegressor.html#sklearn.ensemble.BaggingRegressor" title="sklearn.ensemble.BaggingRegressor"><tt class="xref py py-class docutils literal"><span class="pre">ensemble.BaggingRegressor</span></tt></a> meta-estimators for ensembling
any kind of base estimator. See the <a class="reference internal" href="modules/ensemble.html#bagging"><em>Bagging</em></a> section of
the user guide for details and examples. By <a class="reference external" href="http://www.montefiore.ulg.ac.be/~glouppe/">Gilles Louppe</a>.</li>
<li>New unsupervised feature selection algorithm
<a class="reference internal" href="modules/generated/sklearn.feature_selection.VarianceThreshold.html#sklearn.feature_selection.VarianceThreshold" title="sklearn.feature_selection.VarianceThreshold"><tt class="xref py py-class docutils literal"><span class="pre">feature_selection.VarianceThreshold</span></tt></a>, by <a class="reference external" href="https://github.com/larsmans">Lars Buitinck</a>.</li>
<li>Added <a class="reference internal" href="modules/generated/sklearn.linear_model.RANSACRegressor.html#sklearn.linear_model.RANSACRegressor" title="sklearn.linear_model.RANSACRegressor"><tt class="xref py py-class docutils literal"><span class="pre">linear_model.RANSACRegressor</span></tt></a> meta-estimator for the robust
fitting of regression models. By Johannes Schönberger.</li>
<li>Added <a class="reference internal" href="modules/generated/sklearn.cluster.AgglomerativeClustering.html#sklearn.cluster.AgglomerativeClustering" title="sklearn.cluster.AgglomerativeClustering"><tt class="xref py py-class docutils literal"><span class="pre">cluster.AgglomerativeClustering</span></tt></a> for hierarchical
agglomerative clustering with average linkage, complete linkage and
ward strategies, by <a class="reference external" href="https://github.com/nellev">Nelle Varoquaux</a> and <a class="reference external" href="http://gael-varoquaux.info">Gael Varoquaux</a>.</li>
<li>Shorthand constructors <a class="reference internal" href="modules/generated/sklearn.pipeline.make_pipeline.html#sklearn.pipeline.make_pipeline" title="sklearn.pipeline.make_pipeline"><tt class="xref py py-func docutils literal"><span class="pre">pipeline.make_pipeline</span></tt></a> and
<a class="reference internal" href="modules/generated/sklearn.pipeline.make_union.html#sklearn.pipeline.make_union" title="sklearn.pipeline.make_union"><tt class="xref py py-func docutils literal"><span class="pre">pipeline.make_union</span></tt></a> were added by <a class="reference external" href="https://github.com/larsmans">Lars Buitinck</a>.</li>
<li>Shuffle option for <a class="reference internal" href="modules/generated/sklearn.cross_validation.StratifiedKFold.html#sklearn.cross_validation.StratifiedKFold" title="sklearn.cross_validation.StratifiedKFold"><tt class="xref py py-class docutils literal"><span class="pre">cross_validation.StratifiedKFold</span></tt></a>.
By <a class="reference external" href="https://github.com/jblackburne">Jeffrey Blackburne</a>.</li>
<li>Incremental learning (<tt class="docutils literal"><span class="pre">partial_fit</span></tt>) for Gaussian Naive Bayes by
Imran Haque.</li>
<li>Added <tt class="docutils literal"><span class="pre">partial_fit</span></tt> to <a class="reference internal" href="modules/generated/sklearn.neural_network.BernoulliRBM.html#sklearn.neural_network.BernoulliRBM" title="sklearn.neural_network.BernoulliRBM"><tt class="xref py py-class docutils literal"><span class="pre">BernoulliRBM</span></tt></a>
By <a class="reference external" href="https://github.com/dsullivan7">Danny Sullivan</a>.</li>
<li>Added <a class="reference internal" href="modules/generated/sklearn.learning_curve.learning_curve.html#sklearn.learning_curve.learning_curve" title="sklearn.learning_curve.learning_curve"><tt class="xref py py-func docutils literal"><span class="pre">learning_curve</span></tt></a> utility to
chart performance with respect to training size. See
<a class="reference internal" href="auto_examples/model_selection/plot_learning_curve.html#example-model-selection-plot-learning-curve-py"><em>Plotting Learning Curves</em></a>. By Alexander Fabisch.</li>
<li>Add positive option in <a class="reference internal" href="modules/generated/sklearn.linear_model.LassoCV.html#sklearn.linear_model.LassoCV" title="sklearn.linear_model.LassoCV"><tt class="xref py py-class docutils literal"><span class="pre">LassoCV</span></tt></a> and
<a class="reference internal" href="modules/generated/sklearn.linear_model.ElasticNetCV.html#sklearn.linear_model.ElasticNetCV" title="sklearn.linear_model.ElasticNetCV"><tt class="xref py py-class docutils literal"><span class="pre">ElasticNetCV</span></tt></a>.
By Brian Wignall and <a class="reference external" href="http://alexandre.gramfort.net">Alexandre Gramfort</a>.</li>
<li>Added <a class="reference internal" href="modules/generated/sklearn.linear_model.MultiTaskElasticNetCV.html#sklearn.linear_model.MultiTaskElasticNetCV" title="sklearn.linear_model.MultiTaskElasticNetCV"><tt class="xref py py-class docutils literal"><span class="pre">linear_model.MultiTaskElasticNetCV</span></tt></a> and
<a class="reference internal" href="modules/generated/sklearn.linear_model.MultiTaskLassoCV.html#sklearn.linear_model.MultiTaskLassoCV" title="sklearn.linear_model.MultiTaskLassoCV"><tt class="xref py py-class docutils literal"><span class="pre">linear_model.MultiTaskLassoCV</span></tt></a>. By <a class="reference external" href="https://manojbits.wordpress.com">Manoj Kumar</a>.</li>
<li>Added <a class="reference internal" href="modules/generated/sklearn.manifold.TSNE.html#sklearn.manifold.TSNE" title="sklearn.manifold.TSNE"><tt class="xref py py-class docutils literal"><span class="pre">manifold.TSNE</span></tt></a>. By Alexander Fabisch.</li>
</ul>
</div></blockquote>
</div>
<div class="section" id="id9">
<h3>Enhancements<a class="headerlink" href="#id9" title="Permalink to this headline">¶</a></h3>
<blockquote>
<div><ul class="simple">
<li>Add sparse input support to <a class="reference internal" href="modules/generated/sklearn.ensemble.AdaBoostClassifier.html#sklearn.ensemble.AdaBoostClassifier" title="sklearn.ensemble.AdaBoostClassifier"><tt class="xref py py-class docutils literal"><span class="pre">ensemble.AdaBoostClassifier</span></tt></a> and
<a class="reference internal" href="modules/generated/sklearn.ensemble.AdaBoostRegressor.html#sklearn.ensemble.AdaBoostRegressor" title="sklearn.ensemble.AdaBoostRegressor"><tt class="xref py py-class docutils literal"><span class="pre">ensemble.AdaBoostRegressor</span></tt></a> meta-estimators.
By <a class="reference external" href="https://github.com/hamsal">Hamzeh Alsalhi</a>.</li>
<li>Memory improvements of decision trees, by <a class="reference external" href="http://www.ajoly.org">Arnaud Joly</a>.</li>
<li>Decision trees can now be built in best-first manner by using <tt class="docutils literal"><span class="pre">max_leaf_nodes</span></tt>
as the stopping criteria. Refactored the tree code to use either a
stack or a priority queue for tree building.
By <a class="reference external" href="http://sites.google.com/site/peterprettenhofer/">Peter Prettenhofer</a> and <a class="reference external" href="http://www.montefiore.ulg.ac.be/~glouppe/">Gilles Louppe</a>.</li>
<li>Decision trees can now be fitted on fortran- and c-style arrays, and
non-continuous arrays without the need to make a copy.
If the input array has a different dtype than <tt class="docutils literal"><span class="pre">np.float32</span></tt>, a fortran-
style copy will be made since fortran-style memory layout has speed
advantages. By <a class="reference external" href="http://sites.google.com/site/peterprettenhofer/">Peter Prettenhofer</a> and <a class="reference external" href="http://www.montefiore.ulg.ac.be/~glouppe/">Gilles Louppe</a>.</li>
<li>Speed improvement of regression trees by optimizing the
the computation of the mean square error criterion. This lead
to speed improvement of the tree, forest and gradient boosting tree
modules. By <a class="reference external" href="http://www.ajoly.org">Arnaud Joly</a></li>
<li>The <tt class="docutils literal"><span class="pre">img_to_graph</span></tt> and <tt class="docutils literal"><span class="pre">grid_tograph</span></tt> functions in
<a class="reference internal" href="modules/classes.html#module-sklearn.feature_extraction.image" title="sklearn.feature_extraction.image"><tt class="xref py py-mod docutils literal"><span class="pre">sklearn.feature_extraction.image</span></tt></a> now return <tt class="docutils literal"><span class="pre">np.ndarray</span></tt>
instead of <tt class="docutils literal"><span class="pre">np.matrix</span></tt> when <tt class="docutils literal"><span class="pre">return_as=np.ndarray</span></tt>. See the
Notes section for more information on compatibility.</li>
<li>Changed the internal storage of decision trees to use a struct array.
This fixed some small bugs, while improving code and providing a small
speed gain. By <a class="reference external" href="http://joelnothman.com">Joel Nothman</a>.</li>
<li>Reduce memory usage and overhead when fitting and predicting with forests
of randomized trees in parallel with <tt class="docutils literal"><span class="pre">n_jobs</span> <span class="pre">!=</span> <span class="pre">1</span></tt> by leveraging new
threading backend of joblib 0.8 and releasing the GIL in the tree fitting
Cython code. By <a class="reference external" href="http://twitter.com/ogrisel">Olivier Grisel</a> and <a class="reference external" href="http://www.montefiore.ulg.ac.be/~glouppe/">Gilles Louppe</a>.</li>
<li>Speed improvement of the <tt class="xref py py-mod docutils literal"><span class="pre">sklearn.ensemble.gradient_boosting</span></tt> module.
By <a class="reference external" href="http://www.montefiore.ulg.ac.be/~glouppe/">Gilles Louppe</a> and <a class="reference external" href="http://sites.google.com/site/peterprettenhofer/">Peter Prettenhofer</a>.</li>
<li>Various enhancements to the <tt class="xref py py-mod docutils literal"><span class="pre">sklearn.ensemble.gradient_boosting</span></tt>
module: a <tt class="docutils literal"><span class="pre">warm_start</span></tt> argument to fit additional trees,
a <tt class="docutils literal"><span class="pre">max_leaf_nodes</span></tt> argument to fit GBM style trees,
a <tt class="docutils literal"><span class="pre">monitor</span></tt> fit argument to inspect the estimator during training, and
refactoring of the verbose code. By <a class="reference external" href="http://sites.google.com/site/peterprettenhofer/">Peter Prettenhofer</a>.</li>
<li>Faster <tt class="xref py py-class docutils literal"><span class="pre">sklearn.ensemble.ExtraTrees</span></tt> by caching feature values.
By <a class="reference external" href="http://www.ajoly.org">Arnaud Joly</a>.</li>
<li>Faster depth-based tree building algorithm such as decision tree,
random forest, extra trees or gradient tree boosting (with depth based
growing strategy) by avoiding trying to split on found constant features
in the sample subset. By <a class="reference external" href="http://www.ajoly.org">Arnaud Joly</a>.</li>
<li>Add <tt class="docutils literal"><span class="pre">min_weight_fraction_leaf</span></tt> pre-pruning parameter to tree-based
methods: the minimum weighted fraction of the input samples required to be
at a leaf node. By <a class="reference external" href="http://noel.dawe.me">Noel Dawe</a>.</li>
<li>Added <a class="reference internal" href="modules/generated/sklearn.metrics.pairwise_distances_argmin_min.html#sklearn.metrics.pairwise_distances_argmin_min" title="sklearn.metrics.pairwise_distances_argmin_min"><tt class="xref py py-func docutils literal"><span class="pre">metrics.pairwise_distances_argmin_min</span></tt></a>, by Philippe Gervais.</li>
<li>Added predict method to <a class="reference internal" href="modules/generated/sklearn.cluster.AffinityPropagation.html#sklearn.cluster.AffinityPropagation" title="sklearn.cluster.AffinityPropagation"><tt class="xref py py-class docutils literal"><span class="pre">cluster.AffinityPropagation</span></tt></a> and
<a class="reference internal" href="modules/generated/sklearn.cluster.MeanShift.html#sklearn.cluster.MeanShift" title="sklearn.cluster.MeanShift"><tt class="xref py py-class docutils literal"><span class="pre">cluster.MeanShift</span></tt></a>, by <a class="reference external" href="http://www.mblondel.org">Mathieu Blondel</a>.</li>
<li>Vector and matrix multiplications have been optimised throughout the
library by <a class="reference external" href="https://github.com/dengemann">Denis Engemann</a>, and <a class="reference external" href="http://alexandre.gramfort.net">Alexandre Gramfort</a>.
In particular, they should take less memory with older NumPy versions
(prior to 1.7.2).</li>
<li>Precision-recall and ROC examples now use train_test_split, and have more
explanation of why these metrics are useful. By <a class="reference external" href="http://kastnerkyle.github.io">Kyle Kastner</a></li>
<li>The training algorithm for <a class="reference internal" href="modules/generated/sklearn.decomposition.NMF.html#sklearn.decomposition.NMF" title="sklearn.decomposition.NMF"><tt class="xref py py-class docutils literal"><span class="pre">decomposition.NMF</span></tt></a> is faster for
sparse matrices and has much lower memory complexity, meaning it will
scale up gracefully to large datasets. By <a class="reference external" href="https://github.com/larsmans">Lars Buitinck</a>.</li>
<li>Added svd_method option with default value to “randomized” to
<a class="reference internal" href="modules/generated/sklearn.decomposition.FactorAnalysis.html#sklearn.decomposition.FactorAnalysis" title="sklearn.decomposition.FactorAnalysis"><tt class="xref py py-class docutils literal"><span class="pre">decomposition.FactorAnalysis</span></tt></a> to save memory and
significantly speedup computation by <a class="reference external" href="https://github.com/dengemann">Denis Engemann</a>, and
<a class="reference external" href="http://alexandre.gramfort.net">Alexandre Gramfort</a>.</li>
<li>Changed <a class="reference internal" href="modules/generated/sklearn.cross_validation.StratifiedKFold.html#sklearn.cross_validation.StratifiedKFold" title="sklearn.cross_validation.StratifiedKFold"><tt class="xref py py-class docutils literal"><span class="pre">cross_validation.StratifiedKFold</span></tt></a> to try and
preserve as much of the original ordering of samples as possible so as
not to hide overfitting on datasets with a non-negligible level of
samples dependency.
By <a class="reference external" href="http://danielnouri.org">Daniel Nouri</a> and <a class="reference external" href="http://twitter.com/ogrisel">Olivier Grisel</a>.</li>
<li>Add multi-output support to <a class="reference internal" href="modules/generated/sklearn.gaussian_process.GaussianProcess.html#sklearn.gaussian_process.GaussianProcess" title="sklearn.gaussian_process.GaussianProcess"><tt class="xref py py-class docutils literal"><span class="pre">gaussian_process.GaussianProcess</span></tt></a>
by John Novak.</li>
<li>Norm computations optimized for NumPy 1.6 and later versions by
<a class="reference external" href="https://github.com/larsmans">Lars Buitinck</a>. In particular, the k-means algorithm no longer
needs a temporary data structure the size of its input.</li>
<li><a class="reference internal" href="modules/generated/sklearn.dummy.DummyClassifier.html#sklearn.dummy.DummyClassifier" title="sklearn.dummy.DummyClassifier"><tt class="xref py py-class docutils literal"><span class="pre">dummy.DummyClassifier</span></tt></a> can now be used to predict a constant
output value. By <a class="reference external" href="https://manojbits.wordpress.com">Manoj Kumar</a>.</li>
<li><a class="reference internal" href="modules/generated/sklearn.dummy.DummyRegressor.html#sklearn.dummy.DummyRegressor" title="sklearn.dummy.DummyRegressor"><tt class="xref py py-class docutils literal"><span class="pre">dummy.DummyRegressor</span></tt></a> has now a strategy parameter which allows
to predict the mean, the median of the training set or a constant
output value. By <a class="reference external" href="https://github.com/maheshakya">Maheshakya Wijewardena</a>.</li>
<li>Multi-label classification output in multilabel indicator format
is now supported by <a class="reference internal" href="modules/generated/sklearn.metrics.roc_auc_score.html#sklearn.metrics.roc_auc_score" title="sklearn.metrics.roc_auc_score"><tt class="xref py py-func docutils literal"><span class="pre">metrics.roc_auc_score</span></tt></a> and
<a class="reference internal" href="modules/generated/sklearn.metrics.average_precision_score.html#sklearn.metrics.average_precision_score" title="sklearn.metrics.average_precision_score"><tt class="xref py py-func docutils literal"><span class="pre">metrics.average_precision_score</span></tt></a> by <a class="reference external" href="http://www.ajoly.org">Arnaud Joly</a>.</li>
<li>Significant performance improvements (more than 100x speedup for
large problems) in <a class="reference internal" href="modules/generated/sklearn.isotonic.IsotonicRegression.html#sklearn.isotonic.IsotonicRegression" title="sklearn.isotonic.IsotonicRegression"><tt class="xref py py-class docutils literal"><span class="pre">isotonic.IsotonicRegression</span></tt></a> by
<a class="reference external" href="http://tullo.ch">Andrew Tulloch</a>.</li>
<li>Speed and memory usage improvements to the SGD algorithm for linear
models: it now uses threads, not separate processes, when <tt class="docutils literal"><span class="pre">n_jobs>1</span></tt>.
By <a class="reference external" href="https://github.com/larsmans">Lars Buitinck</a>.</li>
<li>Grid search and cross validation allow NaNs in the input arrays so that
preprocessors such as <a class="reference internal" href="modules/generated/sklearn.preprocessing.Imputer.html#sklearn.preprocessing.Imputer" title="sklearn.preprocessing.Imputer"><tt class="xref py py-class docutils literal"><span class="pre">preprocessing.Imputer</span></tt></a> can be trained within the cross validation loop,
avoiding potentially skewed results.</li>
<li>Ridge regression can now deal with sample weights in feature space
(only sample space until then). By <a class="reference external" href="https://github.com/eickenberg">Michael Eickenberg</a>.
Both solutions are provided by the Cholesky solver.</li>
<li>Several classification and regression metrics now support weighted
samples with the new <tt class="docutils literal"><span class="pre">sample_weight</span></tt> argument:
<a class="reference internal" href="modules/generated/sklearn.metrics.accuracy_score.html#sklearn.metrics.accuracy_score" title="sklearn.metrics.accuracy_score"><tt class="xref py py-func docutils literal"><span class="pre">metrics.accuracy_score</span></tt></a>,
<a class="reference internal" href="modules/generated/sklearn.metrics.zero_one_loss.html#sklearn.metrics.zero_one_loss" title="sklearn.metrics.zero_one_loss"><tt class="xref py py-func docutils literal"><span class="pre">metrics.zero_one_loss</span></tt></a>,
<a class="reference internal" href="modules/generated/sklearn.metrics.precision_score.html#sklearn.metrics.precision_score" title="sklearn.metrics.precision_score"><tt class="xref py py-func docutils literal"><span class="pre">metrics.precision_score</span></tt></a>,
<a class="reference internal" href="modules/generated/sklearn.metrics.average_precision_score.html#sklearn.metrics.average_precision_score" title="sklearn.metrics.average_precision_score"><tt class="xref py py-func docutils literal"><span class="pre">metrics.average_precision_score</span></tt></a>,
<a class="reference internal" href="modules/generated/sklearn.metrics.f1_score.html#sklearn.metrics.f1_score" title="sklearn.metrics.f1_score"><tt class="xref py py-func docutils literal"><span class="pre">metrics.f1_score</span></tt></a>,
<a class="reference internal" href="modules/generated/sklearn.metrics.fbeta_score.html#sklearn.metrics.fbeta_score" title="sklearn.metrics.fbeta_score"><tt class="xref py py-func docutils literal"><span class="pre">metrics.fbeta_score</span></tt></a>,
<a class="reference internal" href="modules/generated/sklearn.metrics.recall_score.html#sklearn.metrics.recall_score" title="sklearn.metrics.recall_score"><tt class="xref py py-func docutils literal"><span class="pre">metrics.recall_score</span></tt></a>,
<a class="reference internal" href="modules/generated/sklearn.metrics.roc_auc_score.html#sklearn.metrics.roc_auc_score" title="sklearn.metrics.roc_auc_score"><tt class="xref py py-func docutils literal"><span class="pre">metrics.roc_auc_score</span></tt></a>,
<a class="reference internal" href="modules/generated/sklearn.metrics.explained_variance_score.html#sklearn.metrics.explained_variance_score" title="sklearn.metrics.explained_variance_score"><tt class="xref py py-func docutils literal"><span class="pre">metrics.explained_variance_score</span></tt></a>,
<a class="reference internal" href="modules/generated/sklearn.metrics.mean_squared_error.html#sklearn.metrics.mean_squared_error" title="sklearn.metrics.mean_squared_error"><tt class="xref py py-func docutils literal"><span class="pre">metrics.mean_squared_error</span></tt></a>,
<a class="reference internal" href="modules/generated/sklearn.metrics.mean_absolute_error.html#sklearn.metrics.mean_absolute_error" title="sklearn.metrics.mean_absolute_error"><tt class="xref py py-func docutils literal"><span class="pre">metrics.mean_absolute_error</span></tt></a>,
<a class="reference internal" href="modules/generated/sklearn.metrics.r2_score.html#sklearn.metrics.r2_score" title="sklearn.metrics.r2_score"><tt class="xref py py-func docutils literal"><span class="pre">metrics.r2_score</span></tt></a>.
By <a class="reference external" href="http://noel.dawe.me">Noel Dawe</a>.</li>
<li>Speed up of the sample generator
<a class="reference internal" href="modules/generated/sklearn.datasets.make_multilabel_classification.html#sklearn.datasets.make_multilabel_classification" title="sklearn.datasets.make_multilabel_classification"><tt class="xref py py-func docutils literal"><span class="pre">datasets.make_multilabel_classification</span></tt></a>. By <a class="reference external" href="http://joelnothman.com">Joel Nothman</a>.</li>
</ul>
</div></blockquote>
</div>
<div class="section" id="id10">
<h3>Documentation improvements<a class="headerlink" href="#id10" title="Permalink to this headline">¶</a></h3>
<blockquote>
<div><ul class="simple">
<li>The <a class="reference internal" href="tutorial/text_analytics/working_with_text_data.html#text-data-tutorial"><em>Working With Text Data</em></a> tutorial
has now been worked in to the main documentation’s tutorial section.
Includes exercises and skeletons for tutorial presentation.
Original tutorial created by several authors including
<a class="reference external" href="http://twitter.com/ogrisel">Olivier Grisel</a>, Lars Buitinck and many others.
Tutorial integration into the scikit-learn documentation
by <a class="reference external" href="https://github.com/jaquesgrobler/scikit-learn/wiki/Jaques-Grobler">Jaques Grobler</a></li>
<li>Added <a class="reference internal" href="modules/computational_performance.html#computational-performance"><em>Computational Performance</em></a>
documentation. Discussion and examples of prediction latency / throughput
and different factors that have influence over speed. Additional tips for
building faster models and choosing a relevant compromise between speed
and predictive power.
By <a class="reference external" href="https://github.com/oddskool">Eustache Diemert</a>.</li>
</ul>
</div></blockquote>
</div>
<div class="section" id="id11">
<h3>Bug fixes<a class="headerlink" href="#id11" title="Permalink to this headline">¶</a></h3>
<blockquote>
<div><ul class="simple">
<li>Fixed bug in <a class="reference internal" href="modules/generated/sklearn.decomposition.MiniBatchDictionaryLearning.html#sklearn.decomposition.MiniBatchDictionaryLearning" title="sklearn.decomposition.MiniBatchDictionaryLearning"><tt class="xref py py-class docutils literal"><span class="pre">decomposition.MiniBatchDictionaryLearning</span></tt></a> :
<tt class="docutils literal"><span class="pre">partial_fit</span></tt> was not working properly.</li>
<li>Fixed bug in <tt class="xref py py-class docutils literal"><span class="pre">linear_model.stochastic_gradient</span></tt> :
<tt class="docutils literal"><span class="pre">l1_ratio</span></tt> was used as <tt class="docutils literal"><span class="pre">(1.0</span> <span class="pre">-</span> <span class="pre">l1_ratio)</span></tt> .</li>
<li>Fixed bug in <a class="reference internal" href="modules/generated/sklearn.multiclass.OneVsOneClassifier.html#sklearn.multiclass.OneVsOneClassifier" title="sklearn.multiclass.OneVsOneClassifier"><tt class="xref py py-class docutils literal"><span class="pre">multiclass.OneVsOneClassifier</span></tt></a> with string
labels</li>
<li>Fixed a bug in <a class="reference internal" href="modules/generated/sklearn.linear_model.LassoCV.html#sklearn.linear_model.LassoCV" title="sklearn.linear_model.LassoCV"><tt class="xref py py-class docutils literal"><span class="pre">LassoCV</span></tt></a> and
<a class="reference internal" href="modules/generated/sklearn.linear_model.ElasticNetCV.html#sklearn.linear_model.ElasticNetCV" title="sklearn.linear_model.ElasticNetCV"><tt class="xref py py-class docutils literal"><span class="pre">ElasticNetCV</span></tt></a>: they would not
pre-compute the Gram matrix with <tt class="docutils literal"><span class="pre">precompute=True</span></tt> or
<tt class="docutils literal"><span class="pre">precompute="auto"</span></tt> and <tt class="docutils literal"><span class="pre">n_samples</span> <span class="pre">></span> <span class="pre">n_features</span></tt>. By <a class="reference external" href="https://manojbits.wordpress.com">Manoj Kumar</a>.</li>
<li>Fixed incorrect estimation of the degrees of freedom in
<a class="reference internal" href="modules/generated/sklearn.feature_selection.f_regression.html#sklearn.feature_selection.f_regression" title="sklearn.feature_selection.f_regression"><tt class="xref py py-func docutils literal"><span class="pre">feature_selection.f_regression</span></tt></a> when variates are not centered.
By <a class="reference external" href="http://parietal.saclay.inria.fr/Members/virgile-fritsch">Virgile Fritsch</a>.</li>
<li>Fixed a race condition in parallel processing with
<tt class="docutils literal"><span class="pre">pre_dispatch</span> <span class="pre">!=</span> <span class="pre">"all"</span></tt> (for instance in <tt class="docutils literal"><span class="pre">cross_val_score</span></tt>).
By <a class="reference external" href="http://twitter.com/ogrisel">Olivier Grisel</a>.</li>
<li>Raise error in <a class="reference internal" href="modules/generated/sklearn.cluster.FeatureAgglomeration.html#sklearn.cluster.FeatureAgglomeration" title="sklearn.cluster.FeatureAgglomeration"><tt class="xref py py-class docutils literal"><span class="pre">cluster.FeatureAgglomeration</span></tt></a> and
<tt class="xref py py-class docutils literal"><span class="pre">cluster.WardAgglomeration</span></tt> when no samples are given,
rather than returning meaningless clustering.</li>
<li>Fixed bug in <tt class="xref py py-class docutils literal"><span class="pre">gradient_boosting.GradientBoostingRegressor</span></tt> with
<tt class="docutils literal"><span class="pre">loss='huber'</span></tt>: <tt class="docutils literal"><span class="pre">gamma</span></tt> might have not been initialized.</li>
<li>Fixed feature importances as computed with a forest of randomized trees
when fit with <tt class="docutils literal"><span class="pre">sample_weight</span> <span class="pre">!=</span> <span class="pre">None</span></tt> and/or with <tt class="docutils literal"><span class="pre">bootstrap=True</span></tt>.
By <a class="reference external" href="http://www.montefiore.ulg.ac.be/~glouppe/">Gilles Louppe</a>.</li>
</ul>
</div></blockquote>
</div>
</div>
<div class="section" id="id12">
<h2>API changes summary<a class="headerlink" href="#id12" title="Permalink to this headline">¶</a></h2>
<blockquote>
<div><ul class="simple">
<li><tt class="xref py py-mod docutils literal"><span class="pre">sklearn.hmm</span></tt> is deprecated. Its removal is planned
for the 0.17 release.</li>
<li>Use of <tt class="xref py py-class docutils literal"><span class="pre">covariance.EllipticEnvelop</span></tt> has now been removed after
deprecation.
Please use <a class="reference internal" href="modules/generated/sklearn.covariance.EllipticEnvelope.html#sklearn.covariance.EllipticEnvelope" title="sklearn.covariance.EllipticEnvelope"><tt class="xref py py-class docutils literal"><span class="pre">covariance.EllipticEnvelope</span></tt></a> instead.</li>
<li><a class="reference internal" href="modules/generated/sklearn.cluster.Ward.html#sklearn.cluster.Ward" title="sklearn.cluster.Ward"><tt class="xref py py-class docutils literal"><span class="pre">cluster.Ward</span></tt></a> is deprecated. Use
<a class="reference internal" href="modules/generated/sklearn.cluster.AgglomerativeClustering.html#sklearn.cluster.AgglomerativeClustering" title="sklearn.cluster.AgglomerativeClustering"><tt class="xref py py-class docutils literal"><span class="pre">cluster.AgglomerativeClustering</span></tt></a> instead.</li>
<li><tt class="xref py py-class docutils literal"><span class="pre">cluster.WardClustering</span></tt> is deprecated. Use</li>
<li><a class="reference internal" href="modules/generated/sklearn.cluster.AgglomerativeClustering.html#sklearn.cluster.AgglomerativeClustering" title="sklearn.cluster.AgglomerativeClustering"><tt class="xref py py-class docutils literal"><span class="pre">cluster.AgglomerativeClustering</span></tt></a> instead.</li>
<li><tt class="xref py py-class docutils literal"><span class="pre">cross_validation.Bootstrap</span></tt> is deprecated.
<a class="reference internal" href="modules/generated/sklearn.cross_validation.KFold.html#sklearn.cross_validation.KFold" title="sklearn.cross_validation.KFold"><tt class="xref py py-class docutils literal"><span class="pre">cross_validation.KFold</span></tt></a> or
<a class="reference internal" href="modules/generated/sklearn.cross_validation.ShuffleSplit.html#sklearn.cross_validation.ShuffleSplit" title="sklearn.cross_validation.ShuffleSplit"><tt class="xref py py-class docutils literal"><span class="pre">cross_validation.ShuffleSplit</span></tt></a> are recommended instead.</li>
<li>Direct support for the sequence of sequences (or list of lists) multilabel
format is deprecated. To convert to and from the supported binary
indicator matrix format, use
<a class="reference internal" href="modules/generated/sklearn.preprocessing.MultiLabelBinarizer.html#sklearn.preprocessing.MultiLabelBinarizer" title="sklearn.preprocessing.MultiLabelBinarizer"><tt class="xref py py-class docutils literal"><span class="pre">MultiLabelBinarizer</span></tt></a>.
By <a class="reference external" href="http://joelnothman.com">Joel Nothman</a>.</li>
<li>Add score method to <a class="reference internal" href="modules/generated/sklearn.decomposition.PCA.html#sklearn.decomposition.PCA" title="sklearn.decomposition.PCA"><tt class="xref py py-class docutils literal"><span class="pre">PCA</span></tt></a> following the model of
probabilistic PCA and deprecate
<tt class="xref py py-class docutils literal"><span class="pre">ProbabilisticPCA</span></tt> model whose
score implementation is not correct. The computation now also exploits the
matrix inversion lemma for faster computation. By <a class="reference external" href="http://alexandre.gramfort.net">Alexandre Gramfort</a>.</li>
<li>The score method of <a class="reference internal" href="modules/generated/sklearn.decomposition.FactorAnalysis.html#sklearn.decomposition.FactorAnalysis" title="sklearn.decomposition.FactorAnalysis"><tt class="xref py py-class docutils literal"><span class="pre">FactorAnalysis</span></tt></a>
now returns the average log-likelihood of the samples. Use score_samples
to get log-likelihood of each sample. By <a class="reference external" href="http://alexandre.gramfort.net">Alexandre Gramfort</a>.</li>
<li>Generating boolean masks (the setting <tt class="docutils literal"><span class="pre">indices=False</span></tt>)
from cross-validation generators is deprecated.
Support for masks will be removed in 0.17.
The generators have produced arrays of indices by default since 0.10.
By <a class="reference external" href="http://joelnothman.com">Joel Nothman</a>.</li>
<li>1-d arrays containing strings with <tt class="docutils literal"><span class="pre">dtype=object</span></tt> (as used in Pandas)
are now considered valid classification targets. This fixes a regression
from version 0.13 in some classifiers. By <a class="reference external" href="http://joelnothman.com">Joel Nothman</a>.</li>
<li>Fix wrong <tt class="docutils literal"><span class="pre">explained_variance_ratio_</span></tt> attribute in
<a class="reference internal" href="modules/generated/sklearn.decomposition.RandomizedPCA.html#sklearn.decomposition.RandomizedPCA" title="sklearn.decomposition.RandomizedPCA"><tt class="xref py py-class docutils literal"><span class="pre">RandomizedPCA</span></tt></a>.
By <a class="reference external" href="http://alexandre.gramfort.net">Alexandre Gramfort</a>.</li>
<li>Fit alphas for each <tt class="docutils literal"><span class="pre">l1_ratio</span></tt> instead of <tt class="docutils literal"><span class="pre">mean_l1_ratio</span></tt> in
<a class="reference internal" href="modules/generated/sklearn.linear_model.ElasticNetCV.html#sklearn.linear_model.ElasticNetCV" title="sklearn.linear_model.ElasticNetCV"><tt class="xref py py-class docutils literal"><span class="pre">linear_model.ElasticNetCV</span></tt></a> and <a class="reference internal" href="modules/generated/sklearn.linear_model.LassoCV.html#sklearn.linear_model.LassoCV" title="sklearn.linear_model.LassoCV"><tt class="xref py py-class docutils literal"><span class="pre">linear_model.LassoCV</span></tt></a>.
This changes the shape of <tt class="docutils literal"><span class="pre">alphas_</span></tt> from <tt class="docutils literal"><span class="pre">(n_alphas,)</span></tt> to
<tt class="docutils literal"><span class="pre">(n_l1_ratio,</span> <span class="pre">n_alphas)</span></tt> if the <tt class="docutils literal"><span class="pre">l1_ratio</span></tt> provided is a 1-D array like
object of length greater than one.
By <a class="reference external" href="https://manojbits.wordpress.com">Manoj Kumar</a>.</li>
<li>Fix <a class="reference internal" href="modules/generated/sklearn.linear_model.ElasticNetCV.html#sklearn.linear_model.ElasticNetCV" title="sklearn.linear_model.ElasticNetCV"><tt class="xref py py-class docutils literal"><span class="pre">linear_model.ElasticNetCV</span></tt></a> and <a class="reference internal" href="modules/generated/sklearn.linear_model.LassoCV.html#sklearn.linear_model.LassoCV" title="sklearn.linear_model.LassoCV"><tt class="xref py py-class docutils literal"><span class="pre">linear_model.LassoCV</span></tt></a>
when fitting intercept and input data is sparse. The automatic grid
of alphas was not computed correctly and the scaling with normalize
was wrong. By <a class="reference external" href="https://manojbits.wordpress.com">Manoj Kumar</a>.</li>
<li>Fix wrong maximal number of features drawn (<tt class="docutils literal"><span class="pre">max_features</span></tt>) at each split
for decision trees, random forests and gradient tree boosting.
Previously, the count for the number of drawn features started only after
one non constant features in the split. This bug fix will affect
computational and generalization performance of those algorithms in the
presence of constant features. To get back previous generalization
performance, you should modify the value of <tt class="docutils literal"><span class="pre">max_features</span></tt>.
By <a class="reference external" href="http://www.ajoly.org">Arnaud Joly</a>.</li>
<li>Fix wrong maximal number of features drawn (<tt class="docutils literal"><span class="pre">max_features</span></tt>) at each split
for <a class="reference internal" href="modules/generated/sklearn.ensemble.ExtraTreesClassifier.html#sklearn.ensemble.ExtraTreesClassifier" title="sklearn.ensemble.ExtraTreesClassifier"><tt class="xref py py-class docutils literal"><span class="pre">ensemble.ExtraTreesClassifier</span></tt></a> and
<a class="reference internal" href="modules/generated/sklearn.ensemble.ExtraTreesRegressor.html#sklearn.ensemble.ExtraTreesRegressor" title="sklearn.ensemble.ExtraTreesRegressor"><tt class="xref py py-class docutils literal"><span class="pre">ensemble.ExtraTreesRegressor</span></tt></a>. Previously, only non constant
features in the split was counted as drawn. Now constant features are
counted as drawn. Furthermore at least one feature must be non constant
in order to make a valid split. This bug fix will affect
computational and generalization performance of extra trees in the
presence of constant features. To get back previous generalization
performance, you should modify the value of <tt class="docutils literal"><span class="pre">max_features</span></tt>.
By <a class="reference external" href="http://www.ajoly.org">Arnaud Joly</a>.</li>
<li>Fix <tt class="xref py py-func docutils literal"><span class="pre">utils.compute_class_weight</span></tt> when <tt class="docutils literal"><span class="pre">class_weight=="auto"</span></tt>.
Previously it was broken for input of non-integer <tt class="docutils literal"><span class="pre">dtype</span></tt> and the
weighted array that was returned was wrong. By <a class="reference external" href="https://manojbits.wordpress.com">Manoj Kumar</a>.</li>
<li>Fix <tt class="xref py py-class docutils literal"><span class="pre">cross_validation.Bootstrap</span></tt> to return <tt class="docutils literal"><span class="pre">ValueError</span></tt>
when <tt class="docutils literal"><span class="pre">n_train</span> <span class="pre">+</span> <span class="pre">n_test</span> <span class="pre">></span> <span class="pre">n</span></tt>. By <a class="reference external" href="https://github.com/rphlypo">Ronald Phlypo</a>.</li>
</ul>
</div></blockquote>
</div>
<div class="section" id="people">
<h2>People<a class="headerlink" href="#people" title="Permalink to this headline">¶</a></h2>
<p>List of contributors for release 0.15 by number of commits.</p>
<ul class="simple">
<li>312 Olivier Grisel</li>
<li>275 Lars Buitinck</li>
<li>221 Gael Varoquaux</li>
<li>148 Arnaud Joly</li>
<li>134 Johannes Schönberger</li>
<li>119 Gilles Louppe</li>
<li>113 Joel Nothman</li>
<li>111 Alexandre Gramfort</li>
<li>95 Jaques Grobler</li>
<li>89 Denis Engemann</li>
<li>83 Peter Prettenhofer</li>
<li>83 Alexander Fabisch</li>
<li>62 Mathieu Blondel</li>
<li>60 Eustache Diemert</li>
<li>60 Nelle Varoquaux</li>
<li>49 Michael Bommarito</li>
<li>45 Manoj-Kumar-S</li>
<li>28 Kyle Kastner</li>
<li>26 Andreas Mueller</li>
<li>22 Noel Dawe</li>
<li>21 Maheshakya Wijewardena</li>
<li>21 Brooke Osborn</li>