-
Notifications
You must be signed in to change notification settings - Fork 81
/
Copy pathfaq.html
736 lines (701 loc) · 60.7 KB
/
faq.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
<!DOCTYPE html>
<!--[if IE 8]><html class="no-js lt-ie9" lang="en" > <![endif]-->
<!--[if gt IE 8]><!--> <html class="no-js" lang="en" > <!--<![endif]-->
<head>
<meta charset="utf-8">
<meta name="generator" content="Docutils 0.19: https://docutils.sourceforge.io/" />
<meta property="og:title" content="Frequently Asked Questions" />
<meta property="og:type" content="website" />
<meta property="og:url" content="https://scikit-learn/stable/faq.html" />
<meta property="og:site_name" content="scikit-learn" />
<meta property="og:description" content="Here we try to give some answers to questions that regularly pop up on the mailing list. Table of Contents: About the project- What is the project name (a lot of people get it wrong)?, How do you p..." />
<meta property="og:image" content="https://scikit-learn.org/stable/_static/scikit-learn-logo-small.png" />
<meta property="og:image:alt" content="scikit-learn" />
<meta name="description" content="Here we try to give some answers to questions that regularly pop up on the mailing list. Table of Contents: About the project- What is the project name (a lot of people get it wrong)?, How do you p..." />
<meta name="viewport" content="width=device-width, initial-scale=1.0">
<title>Frequently Asked Questions — scikit-learn 1.4.dev0 documentation</title>
<link rel="canonical" href="http://scikit-learn.org/stable/faq.html" />
<link rel="shortcut icon" href="_static/favicon.ico"/>
<link rel="stylesheet" href="_static/css/vendor/bootstrap.min.css" type="text/css" />
<link rel="stylesheet" href="_static/pygments.css" type="text/css" />
<link rel="stylesheet" href="_static/css/theme.css" type="text/css" />
<link rel="stylesheet" href="_static/copybutton.css" type="text/css" />
<link rel="stylesheet" href="_static/plot_directive.css" type="text/css" />
<link rel="stylesheet" href="https://fonts.googleapis.com/css?family=Vibur" type="text/css" />
<link rel="stylesheet" href="_static/jupyterlite_sphinx.css" type="text/css" />
<link rel="stylesheet" href="_static/sg_gallery.css" type="text/css" />
<link rel="stylesheet" href="_static/sg_gallery-binder.css" type="text/css" />
<link rel="stylesheet" href="_static/sg_gallery-dataframe.css" type="text/css" />
<link rel="stylesheet" href="_static/sg_gallery-rendered-html.css" type="text/css" />
<link rel="stylesheet" href="_static/css/theme.css" type="text/css" />
<script id="documentation_options" data-url_root="./" src="_static/documentation_options.js"></script>
<script src="_static/js/vendor/jquery-3.6.3.slim.min.js"></script>
</head>
<body>
<nav id="navbar" class="sk-docs-navbar navbar navbar-expand-md navbar-light bg-light py-0">
<div class="container-fluid sk-docs-container px-0">
<a class="navbar-brand py-0" href="index.html">
<img
class="sk-brand-img"
src="_static/scikit-learn-logo-small.png"
alt="logo"/>
</a>
<button
id="sk-navbar-toggler"
class="navbar-toggler"
type="button"
data-toggle="collapse"
data-target="#navbarSupportedContent"
aria-controls="navbarSupportedContent"
aria-expanded="false"
aria-label="Toggle navigation"
>
<span class="navbar-toggler-icon"></span>
</button>
<div class="sk-navbar-collapse collapse navbar-collapse" id="navbarSupportedContent">
<ul class="navbar-nav mr-auto">
<li class="nav-item">
<a class="sk-nav-link nav-link" href="install.html">Install</a>
</li>
<li class="nav-item">
<a class="sk-nav-link nav-link" href="user_guide.html">User Guide</a>
</li>
<li class="nav-item">
<a class="sk-nav-link nav-link" href="modules/classes.html">API</a>
</li>
<li class="nav-item">
<a class="sk-nav-link nav-link" href="auto_examples/index.html">Examples</a>
</li>
<li class="nav-item">
<a class="sk-nav-link nav-link" target="_blank" rel="noopener noreferrer" href="https://blog.scikit-learn.org/">Community</a>
</li>
<li class="nav-item">
<a class="sk-nav-link nav-link nav-more-item-mobile-items" href="getting_started.html" >Getting Started</a>
</li>
<li class="nav-item">
<a class="sk-nav-link nav-link nav-more-item-mobile-items" href="tutorial/index.html" >Tutorial</a>
</li>
<li class="nav-item">
<a class="sk-nav-link nav-link nav-more-item-mobile-items" href="whats_new/v1.4.html" >What's new</a>
</li>
<li class="nav-item">
<a class="sk-nav-link nav-link nav-more-item-mobile-items" href="glossary.html" >Glossary</a>
</li>
<li class="nav-item">
<a class="sk-nav-link nav-link nav-more-item-mobile-items" href="developers/index.html" >Development</a>
</li>
<li class="nav-item">
<a class="sk-nav-link nav-link nav-more-item-mobile-items" href="#" >FAQ</a>
</li>
<li class="nav-item">
<a class="sk-nav-link nav-link nav-more-item-mobile-items" href="support.html" >Support</a>
</li>
<li class="nav-item">
<a class="sk-nav-link nav-link nav-more-item-mobile-items" href="related_projects.html" >Related packages</a>
</li>
<li class="nav-item">
<a class="sk-nav-link nav-link nav-more-item-mobile-items" href="roadmap.html" >Roadmap</a>
</li>
<li class="nav-item">
<a class="sk-nav-link nav-link nav-more-item-mobile-items" href="governance.html" >Governance</a>
</li>
<li class="nav-item">
<a class="sk-nav-link nav-link nav-more-item-mobile-items" href="about.html" >About us</a>
</li>
<li class="nav-item">
<a class="sk-nav-link nav-link nav-more-item-mobile-items" href="https://github.com/scikit-learn/scikit-learn" >GitHub</a>
</li>
<li class="nav-item">
<a class="sk-nav-link nav-link nav-more-item-mobile-items" href="https://scikit-learn.org/dev/versions.html" >Other Versions and Download</a>
</li>
<li class="nav-item dropdown nav-more-item-dropdown">
<a class="sk-nav-link nav-link dropdown-toggle" href="#" id="navbarDropdown" role="button" data-toggle="dropdown" aria-haspopup="true" aria-expanded="false">More</a>
<div class="dropdown-menu" aria-labelledby="navbarDropdown">
<a class="sk-nav-dropdown-item dropdown-item" href="getting_started.html" >Getting Started</a>
<a class="sk-nav-dropdown-item dropdown-item" href="tutorial/index.html" >Tutorial</a>
<a class="sk-nav-dropdown-item dropdown-item" href="whats_new/v1.4.html" >What's new</a>
<a class="sk-nav-dropdown-item dropdown-item" href="glossary.html" >Glossary</a>
<a class="sk-nav-dropdown-item dropdown-item" href="developers/index.html" >Development</a>
<a class="sk-nav-dropdown-item dropdown-item" href="#" >FAQ</a>
<a class="sk-nav-dropdown-item dropdown-item" href="support.html" >Support</a>
<a class="sk-nav-dropdown-item dropdown-item" href="related_projects.html" >Related packages</a>
<a class="sk-nav-dropdown-item dropdown-item" href="roadmap.html" >Roadmap</a>
<a class="sk-nav-dropdown-item dropdown-item" href="governance.html" >Governance</a>
<a class="sk-nav-dropdown-item dropdown-item" href="about.html" >About us</a>
<a class="sk-nav-dropdown-item dropdown-item" href="https://github.com/scikit-learn/scikit-learn" >GitHub</a>
<a class="sk-nav-dropdown-item dropdown-item" href="https://scikit-learn.org/dev/versions.html" >Other Versions and Download</a>
</div>
</li>
</ul>
<div id="searchbox" role="search">
<div class="searchformwrapper">
<form class="search" action="search.html" method="get">
<input class="sk-search-text-input" type="text" name="q" aria-labelledby="searchlabel" />
<input class="sk-search-text-btn" type="submit" value="Go" />
</form>
</div>
</div>
</div>
</div>
</nav>
<div class="d-flex" id="sk-doc-wrapper">
<input type="checkbox" name="sk-toggle-checkbox" id="sk-toggle-checkbox">
<label id="sk-sidemenu-toggle" class="sk-btn-toggle-toc btn sk-btn-primary" for="sk-toggle-checkbox">Toggle Menu</label>
<div id="sk-sidebar-wrapper" class="border-right">
<div class="sk-sidebar-toc-wrapper">
<div class="btn-group w-100 mb-2" role="group" aria-label="rellinks">
<a href="install.html" role="button" class="btn sk-btn-rellink py-1" sk-rellink-tooltip="Installing scikit-learn">Prev</a><a href="preface.html" role="button" class="btn sk-btn-rellink py-1" sk-rellink-tooltip="Welcome to scikit-learn">Up</a>
<a href="support.html" role="button" class="btn sk-btn-rellink py-1" sk-rellink-tooltip="Support">Next</a>
</div>
<div class="alert alert-danger p-1 mb-2" role="alert">
<p class="text-center mb-0">
<strong>scikit-learn 1.4.dev0</strong><br/>
<a href="http://scikit-learn.org/dev/versions.html">Other versions</a>
</p>
</div>
<div class="alert alert-warning p-1 mb-2" role="alert">
<p class="text-center mb-0">
Please <a class="font-weight-bold" href="about.html#citing-scikit-learn"><string>cite us</string></a> if you use the software.
</p>
</div>
<div class="sk-sidebar-toc">
<ul>
<li><a class="reference internal" href="#">Frequently Asked Questions</a><ul>
<li><a class="reference internal" href="#about-the-project">About the project</a><ul>
<li><a class="reference internal" href="#what-is-the-project-name-a-lot-of-people-get-it-wrong">What is the project name (a lot of people get it wrong)?</a></li>
<li><a class="reference internal" href="#how-do-you-pronounce-the-project-name">How do you pronounce the project name?</a></li>
<li><a class="reference internal" href="#why-scikit">Why scikit?</a></li>
<li><a class="reference internal" href="#do-you-support-pypy">Do you support PyPy?</a></li>
</ul>
</li>
<li><a class="reference internal" href="#implementation-decisions">Implementation decisions</a><ul>
<li><a class="reference internal" href="#why-is-there-no-support-for-deep-or-reinforcement-learning-will-there-be-support-for-deep-or-reinforcement-learning-in-scikit-learn">Why is there no support for deep or reinforcement learning / Will there be support for deep or reinforcement learning in scikit-learn?</a></li>
<li><a class="reference internal" href="#will-you-add-graphical-models-or-sequence-prediction-to-scikit-learn">Will you add graphical models or sequence prediction to scikit-learn?</a></li>
<li><a class="reference internal" href="#why-did-you-remove-hmms-from-scikit-learn">Why did you remove HMMs from scikit-learn?</a></li>
<li><a class="reference internal" href="#will-you-add-gpu-support">Will you add GPU support?</a></li>
<li><a class="reference internal" href="#why-do-categorical-variables-need-preprocessing-in-scikit-learn-compared-to-other-tools">Why do categorical variables need preprocessing in scikit-learn, compared to other tools?</a></li>
<li><a class="reference internal" href="#why-does-scikit-learn-not-directly-work-with-for-example-pandas-dataframe">Why does Scikit-learn not directly work with, for example, pandas.DataFrame?</a></li>
<li><a class="reference internal" href="#do-you-plan-to-implement-transform-for-target-y-in-a-pipeline">Do you plan to implement transform for target y in a pipeline?</a></li>
<li><a class="reference internal" href="#why-are-there-so-many-different-estimators-for-linear-models">Why are there so many different estimators for linear models?</a></li>
</ul>
</li>
<li><a class="reference internal" href="#contributing">Contributing</a><ul>
<li><a class="reference internal" href="#how-can-i-contribute-to-scikit-learn">How can I contribute to scikit-learn?</a></li>
<li><a class="reference internal" href="#why-is-my-pull-request-not-getting-any-attention">Why is my pull request not getting any attention?</a></li>
<li><a class="reference internal" href="#what-are-the-inclusion-criteria-for-new-algorithms">What are the inclusion criteria for new algorithms ?</a></li>
<li><a class="reference internal" href="#why-are-you-so-selective-on-what-algorithms-you-include-in-scikit-learn">Why are you so selective on what algorithms you include in scikit-learn?</a></li>
</ul>
</li>
<li><a class="reference internal" href="#using-scikit-learn">Using scikit-learn</a><ul>
<li><a class="reference internal" href="#what-s-the-best-way-to-get-help-on-scikit-learn-usage">What’s the best way to get help on scikit-learn usage?</a></li>
<li><a class="reference internal" href="#how-should-i-save-export-or-deploy-estimators-for-production">How should I save, export or deploy estimators for production?</a></li>
<li><a class="reference internal" href="#how-can-i-create-a-bunch-object">How can I create a bunch object?</a></li>
<li><a class="reference internal" href="#how-can-i-load-my-own-datasets-into-a-format-usable-by-scikit-learn">How can I load my own datasets into a format usable by scikit-learn?</a></li>
<li><a class="reference internal" href="#how-do-i-deal-with-string-data-or-trees-graphs">How do I deal with string data (or trees, graphs…)?</a></li>
<li><a class="reference internal" href="#why-do-i-sometime-get-a-crash-freeze-with-n-jobs-1-under-osx-or-linux">Why do I sometime get a crash/freeze with n_jobs > 1 under OSX or Linux?</a></li>
<li><a class="reference internal" href="#why-does-my-job-use-more-cores-than-specified-with-n-jobs">Why does my job use more cores than specified with n_jobs?</a></li>
<li><a class="reference internal" href="#how-do-i-set-a-random-state-for-an-entire-execution">How do I set a <code class="docutils literal notranslate"><span class="pre">random_state</span></code> for an entire execution?</a></li>
</ul>
</li>
</ul>
</li>
</ul>
</div>
</div>
</div>
<div id="sk-page-content-wrapper">
<div class="sk-page-content container-fluid body px-md-3" role="main">
<section id="frequently-asked-questions">
<span id="faq"></span><h1>Frequently Asked Questions<a class="headerlink" href="#frequently-asked-questions" title="Permalink to this heading">¶</a></h1>
<p>Here we try to give some answers to questions that regularly pop up on the mailing list.</p>
<nav class="contents local" id="table-of-contents">
<p class="topic-title">Table of Contents</p>
<ul class="simple">
<li><p><a class="reference internal" href="#about-the-project" id="id2">About the project</a></p>
<ul>
<li><p><a class="reference internal" href="#what-is-the-project-name-a-lot-of-people-get-it-wrong" id="id3">What is the project name (a lot of people get it wrong)?</a></p></li>
<li><p><a class="reference internal" href="#how-do-you-pronounce-the-project-name" id="id4">How do you pronounce the project name?</a></p></li>
<li><p><a class="reference internal" href="#why-scikit" id="id5">Why scikit?</a></p></li>
<li><p><a class="reference internal" href="#do-you-support-pypy" id="id6">Do you support PyPy?</a></p></li>
</ul>
</li>
<li><p><a class="reference internal" href="#implementation-decisions" id="id7">Implementation decisions</a></p>
<ul>
<li><p><a class="reference internal" href="#why-is-there-no-support-for-deep-or-reinforcement-learning-will-there-be-support-for-deep-or-reinforcement-learning-in-scikit-learn" id="id8">Why is there no support for deep or reinforcement learning / Will there be support for deep or reinforcement learning in scikit-learn?</a></p></li>
<li><p><a class="reference internal" href="#will-you-add-graphical-models-or-sequence-prediction-to-scikit-learn" id="id9">Will you add graphical models or sequence prediction to scikit-learn?</a></p></li>
<li><p><a class="reference internal" href="#why-did-you-remove-hmms-from-scikit-learn" id="id10">Why did you remove HMMs from scikit-learn?</a></p></li>
<li><p><a class="reference internal" href="#will-you-add-gpu-support" id="id11">Will you add GPU support?</a></p></li>
<li><p><a class="reference internal" href="#why-do-categorical-variables-need-preprocessing-in-scikit-learn-compared-to-other-tools" id="id12">Why do categorical variables need preprocessing in scikit-learn, compared to other tools?</a></p></li>
<li><p><a class="reference internal" href="#why-does-scikit-learn-not-directly-work-with-for-example-pandas-dataframe" id="id13">Why does Scikit-learn not directly work with, for example, pandas.DataFrame?</a></p></li>
<li><p><a class="reference internal" href="#do-you-plan-to-implement-transform-for-target-y-in-a-pipeline" id="id14">Do you plan to implement transform for target y in a pipeline?</a></p></li>
<li><p><a class="reference internal" href="#why-are-there-so-many-different-estimators-for-linear-models" id="id15">Why are there so many different estimators for linear models?</a></p></li>
</ul>
</li>
<li><p><a class="reference internal" href="#contributing" id="id16">Contributing</a></p>
<ul>
<li><p><a class="reference internal" href="#how-can-i-contribute-to-scikit-learn" id="id17">How can I contribute to scikit-learn?</a></p></li>
<li><p><a class="reference internal" href="#why-is-my-pull-request-not-getting-any-attention" id="id18">Why is my pull request not getting any attention?</a></p></li>
<li><p><a class="reference internal" href="#what-are-the-inclusion-criteria-for-new-algorithms" id="id19">What are the inclusion criteria for new algorithms ?</a></p></li>
<li><p><a class="reference internal" href="#why-are-you-so-selective-on-what-algorithms-you-include-in-scikit-learn" id="id20">Why are you so selective on what algorithms you include in scikit-learn?</a></p></li>
</ul>
</li>
<li><p><a class="reference internal" href="#using-scikit-learn" id="id21">Using scikit-learn</a></p>
<ul>
<li><p><a class="reference internal" href="#what-s-the-best-way-to-get-help-on-scikit-learn-usage" id="id22">What’s the best way to get help on scikit-learn usage?</a></p></li>
<li><p><a class="reference internal" href="#how-should-i-save-export-or-deploy-estimators-for-production" id="id23">How should I save, export or deploy estimators for production?</a></p></li>
<li><p><a class="reference internal" href="#how-can-i-create-a-bunch-object" id="id24">How can I create a bunch object?</a></p></li>
<li><p><a class="reference internal" href="#how-can-i-load-my-own-datasets-into-a-format-usable-by-scikit-learn" id="id25">How can I load my own datasets into a format usable by scikit-learn?</a></p></li>
<li><p><a class="reference internal" href="#how-do-i-deal-with-string-data-or-trees-graphs" id="id26">How do I deal with string data (or trees, graphs…)?</a></p></li>
<li><p><a class="reference internal" href="#why-do-i-sometime-get-a-crash-freeze-with-n-jobs-1-under-osx-or-linux" id="id27">Why do I sometime get a crash/freeze with n_jobs > 1 under OSX or Linux?</a></p></li>
<li><p><a class="reference internal" href="#why-does-my-job-use-more-cores-than-specified-with-n-jobs" id="id28">Why does my job use more cores than specified with n_jobs?</a></p></li>
<li><p><a class="reference internal" href="#how-do-i-set-a-random-state-for-an-entire-execution" id="id29">How do I set a <code class="docutils literal notranslate"><span class="pre">random_state</span></code> for an entire execution?</a></p></li>
</ul>
</li>
</ul>
</nav>
<section id="about-the-project">
<h2><a class="toc-backref" href="#id2" role="doc-backlink">About the project</a><a class="headerlink" href="#about-the-project" title="Permalink to this heading">¶</a></h2>
<section id="what-is-the-project-name-a-lot-of-people-get-it-wrong">
<h3><a class="toc-backref" href="#id3" role="doc-backlink">What is the project name (a lot of people get it wrong)?</a><a class="headerlink" href="#what-is-the-project-name-a-lot-of-people-get-it-wrong" title="Permalink to this heading">¶</a></h3>
<p>scikit-learn, but not scikit or SciKit nor sci-kit learn.
Also not scikits.learn or scikits-learn, which were previously used.</p>
</section>
<section id="how-do-you-pronounce-the-project-name">
<h3><a class="toc-backref" href="#id4" role="doc-backlink">How do you pronounce the project name?</a><a class="headerlink" href="#how-do-you-pronounce-the-project-name" title="Permalink to this heading">¶</a></h3>
<p>sy-kit learn. sci stands for science!</p>
</section>
<section id="why-scikit">
<h3><a class="toc-backref" href="#id5" role="doc-backlink">Why scikit?</a><a class="headerlink" href="#why-scikit" title="Permalink to this heading">¶</a></h3>
<p>There are multiple scikits, which are scientific toolboxes built around SciPy.
Apart from scikit-learn, another popular one is <a class="reference external" href="https://scikit-image.org/">scikit-image</a>.</p>
</section>
<section id="do-you-support-pypy">
<h3><a class="toc-backref" href="#id6" role="doc-backlink">Do you support PyPy?</a><a class="headerlink" href="#do-you-support-pypy" title="Permalink to this heading">¶</a></h3>
<p>scikit-learn is regularly tested and maintained to work with
<a class="reference external" href="https://pypy.org/">PyPy</a> (an alternative Python implementation with
a built-in just-in-time compiler).</p>
<p>Note however that this support is still considered experimental and specific
components might behave slightly differently. Please refer to the test
suite of the specific module of interest for more details.</p>
</section>
</section>
<section id="implementation-decisions">
<h2><a class="toc-backref" href="#id7" role="doc-backlink">Implementation decisions</a><a class="headerlink" href="#implementation-decisions" title="Permalink to this heading">¶</a></h2>
<section id="why-is-there-no-support-for-deep-or-reinforcement-learning-will-there-be-support-for-deep-or-reinforcement-learning-in-scikit-learn">
<h3><a class="toc-backref" href="#id8" role="doc-backlink">Why is there no support for deep or reinforcement learning / Will there be support for deep or reinforcement learning in scikit-learn?</a><a class="headerlink" href="#why-is-there-no-support-for-deep-or-reinforcement-learning-will-there-be-support-for-deep-or-reinforcement-learning-in-scikit-learn" title="Permalink to this heading">¶</a></h3>
<p>Deep learning and reinforcement learning both require a rich vocabulary to
define an architecture, with deep learning additionally requiring
GPUs for efficient computing. However, neither of these fit within
the design constraints of scikit-learn; as a result, deep learning
and reinforcement learning are currently out of scope for what
scikit-learn seeks to achieve.</p>
<p>You can find more information about addition of gpu support at
<a class="reference internal" href="#will-you-add-gpu-support">Will you add GPU support?</a>.</p>
<p>Note that scikit-learn currently implements a simple multilayer perceptron
in <a class="reference internal" href="modules/classes.html#module-sklearn.neural_network" title="sklearn.neural_network"><code class="xref py py-mod docutils literal notranslate"><span class="pre">sklearn.neural_network</span></code></a>. We will only accept bug fixes for this module.
If you want to implement more complex deep learning models, please turn to
popular deep learning frameworks such as
<a class="reference external" href="https://www.tensorflow.org/">tensorflow</a>,
<a class="reference external" href="https://keras.io/">keras</a>
and <a class="reference external" href="https://pytorch.org/">pytorch</a>.</p>
</section>
<section id="will-you-add-graphical-models-or-sequence-prediction-to-scikit-learn">
<span id="adding-graphical-models"></span><h3><a class="toc-backref" href="#id9" role="doc-backlink">Will you add graphical models or sequence prediction to scikit-learn?</a><a class="headerlink" href="#will-you-add-graphical-models-or-sequence-prediction-to-scikit-learn" title="Permalink to this heading">¶</a></h3>
<p>Not in the foreseeable future.
scikit-learn tries to provide a unified API for the basic tasks in machine
learning, with pipelines and meta-algorithms like grid search to tie
everything together. The required concepts, APIs, algorithms and
expertise required for structured learning are different from what
scikit-learn has to offer. If we started doing arbitrary structured
learning, we’d need to redesign the whole package and the project
would likely collapse under its own weight.</p>
<p>There are two projects with API similar to scikit-learn that
do structured prediction:</p>
<ul class="simple">
<li><p><a class="reference external" href="https://pystruct.github.io/">pystruct</a> handles general structured
learning (focuses on SSVMs on arbitrary graph structures with
approximate inference; defines the notion of sample as an instance of
the graph structure)</p></li>
<li><p><a class="reference external" href="https://larsmans.github.io/seqlearn/">seqlearn</a> handles sequences only
(focuses on exact inference; has HMMs, but mostly for the sake of
completeness; treats a feature vector as a sample and uses an offset encoding
for the dependencies between feature vectors)</p></li>
</ul>
</section>
<section id="why-did-you-remove-hmms-from-scikit-learn">
<h3><a class="toc-backref" href="#id10" role="doc-backlink">Why did you remove HMMs from scikit-learn?</a><a class="headerlink" href="#why-did-you-remove-hmms-from-scikit-learn" title="Permalink to this heading">¶</a></h3>
<p>See <a class="reference internal" href="#adding-graphical-models"><span class="std std-ref">Will you add graphical models or sequence prediction to scikit-learn?</span></a>.</p>
</section>
<section id="will-you-add-gpu-support">
<h3><a class="toc-backref" href="#id11" role="doc-backlink">Will you add GPU support?</a><a class="headerlink" href="#will-you-add-gpu-support" title="Permalink to this heading">¶</a></h3>
<p>No, or at least not in the near future. The main reason is that GPU support
will introduce many software dependencies and introduce platform specific
issues. scikit-learn is designed to be easy to install on a wide variety of
platforms. Outside of neural networks, GPUs don’t play a large role in machine
learning today, and much larger gains in speed can often be achieved by a
careful choice of algorithms.</p>
</section>
<section id="why-do-categorical-variables-need-preprocessing-in-scikit-learn-compared-to-other-tools">
<h3><a class="toc-backref" href="#id12" role="doc-backlink">Why do categorical variables need preprocessing in scikit-learn, compared to other tools?</a><a class="headerlink" href="#why-do-categorical-variables-need-preprocessing-in-scikit-learn-compared-to-other-tools" title="Permalink to this heading">¶</a></h3>
<p>Most of scikit-learn assumes data is in NumPy arrays or SciPy sparse matrices
of a single numeric dtype. These do not explicitly represent categorical
variables at present. Thus, unlike R’s data.frames or pandas.DataFrame, we
require explicit conversion of categorical features to numeric values, as
discussed in <a class="reference internal" href="modules/preprocessing.html#preprocessing-categorical-features"><span class="std std-ref">Encoding categorical features</span></a>.
See also <a class="reference internal" href="auto_examples/compose/plot_column_transformer_mixed_types.html#sphx-glr-auto-examples-compose-plot-column-transformer-mixed-types-py"><span class="std std-ref">Column Transformer with Mixed Types</span></a> for an
example of working with heterogeneous (e.g. categorical and numeric) data.</p>
</section>
<section id="why-does-scikit-learn-not-directly-work-with-for-example-pandas-dataframe">
<h3><a class="toc-backref" href="#id13" role="doc-backlink">Why does Scikit-learn not directly work with, for example, pandas.DataFrame?</a><a class="headerlink" href="#why-does-scikit-learn-not-directly-work-with-for-example-pandas-dataframe" title="Permalink to this heading">¶</a></h3>
<p>The homogeneous NumPy and SciPy data objects currently expected are most
efficient to process for most operations. Extensive work would also be needed
to support Pandas categorical types. Restricting input to homogeneous
types therefore reduces maintenance cost and encourages usage of efficient
data structures.</p>
<p>Note however that <a class="reference internal" href="modules/generated/sklearn.compose.ColumnTransformer.html#sklearn.compose.ColumnTransformer" title="sklearn.compose.ColumnTransformer"><code class="xref py py-class docutils literal notranslate"><span class="pre">ColumnTransformer</span></code></a> makes it
convenient to handle heterogeneous pandas dataframes by mapping homogeneous subsets of
dataframe columns selected by name or dtype to dedicated scikit-learn transformers.</p>
<p>Therefore <a class="reference internal" href="modules/generated/sklearn.compose.ColumnTransformer.html#sklearn.compose.ColumnTransformer" title="sklearn.compose.ColumnTransformer"><code class="xref py py-class docutils literal notranslate"><span class="pre">ColumnTransformer</span></code></a> are often used in the first
step of scikit-learn pipelines when dealing
with heterogeneous dataframes (see <a class="reference internal" href="modules/compose.html#pipeline"><span class="std std-ref">Pipeline: chaining estimators</span></a> for more details).</p>
<p>See also <a class="reference internal" href="auto_examples/compose/plot_column_transformer_mixed_types.html#sphx-glr-auto-examples-compose-plot-column-transformer-mixed-types-py"><span class="std std-ref">Column Transformer with Mixed Types</span></a>
for an example of working with heterogeneous (e.g. categorical and numeric) data.</p>
</section>
<section id="do-you-plan-to-implement-transform-for-target-y-in-a-pipeline">
<h3><a class="toc-backref" href="#id14" role="doc-backlink">Do you plan to implement transform for target y in a pipeline?</a><a class="headerlink" href="#do-you-plan-to-implement-transform-for-target-y-in-a-pipeline" title="Permalink to this heading">¶</a></h3>
<p>Currently transform only works for features X in a pipeline.
There’s a long-standing discussion about
not being able to transform y in a pipeline.
Follow on github issue
<a class="reference external" href="https://github.com/scikit-learn/scikit-learn/issues/4143">#4143</a>.
Meanwhile check out
<a class="reference internal" href="modules/generated/sklearn.compose.TransformedTargetRegressor.html#sklearn.compose.TransformedTargetRegressor" title="sklearn.compose.TransformedTargetRegressor"><code class="xref py py-class docutils literal notranslate"><span class="pre">TransformedTargetRegressor</span></code></a>,
<a class="reference external" href="https://github.com/mcasl/PipeGraph">pipegraph</a>,
<a class="reference external" href="https://github.com/scikit-learn-contrib/imbalanced-learn">imbalanced-learn</a>.
Note that Scikit-learn solved for the case where y
has an invertible transformation applied before training
and inverted after prediction. Scikit-learn intends to solve for
use cases where y should be transformed at training time
and not at test time, for resampling and similar uses,
like at <code class="docutils literal notranslate"><span class="pre">imbalanced-learn</span></code>.
In general, these use cases can be solved
with a custom meta estimator rather than a Pipeline</p>
</section>
<section id="why-are-there-so-many-different-estimators-for-linear-models">
<h3><a class="toc-backref" href="#id15" role="doc-backlink">Why are there so many different estimators for linear models?</a><a class="headerlink" href="#why-are-there-so-many-different-estimators-for-linear-models" title="Permalink to this heading">¶</a></h3>
<p>Usually, there is one classifier and one regressor per model type, e.g.
<a class="reference internal" href="modules/generated/sklearn.ensemble.GradientBoostingClassifier.html#sklearn.ensemble.GradientBoostingClassifier" title="sklearn.ensemble.GradientBoostingClassifier"><code class="xref py py-class docutils literal notranslate"><span class="pre">GradientBoostingClassifier</span></code></a> and
<a class="reference internal" href="modules/generated/sklearn.ensemble.GradientBoostingRegressor.html#sklearn.ensemble.GradientBoostingRegressor" title="sklearn.ensemble.GradientBoostingRegressor"><code class="xref py py-class docutils literal notranslate"><span class="pre">GradientBoostingRegressor</span></code></a>. Both have similar options and
both have the parameter <code class="docutils literal notranslate"><span class="pre">loss</span></code>, which is especially useful in the regression
case as it enables the estimation of conditional mean as well as conditional
quantiles.</p>
<p>For linear models, there are many estimator classes which are very close to
each other. Let us have a look at</p>
<ul class="simple">
<li><p><a class="reference internal" href="modules/generated/sklearn.linear_model.LinearRegression.html#sklearn.linear_model.LinearRegression" title="sklearn.linear_model.LinearRegression"><code class="xref py py-class docutils literal notranslate"><span class="pre">LinearRegression</span></code></a>, no penalty</p></li>
<li><p><a class="reference internal" href="modules/generated/sklearn.linear_model.Ridge.html#sklearn.linear_model.Ridge" title="sklearn.linear_model.Ridge"><code class="xref py py-class docutils literal notranslate"><span class="pre">Ridge</span></code></a>, L2 penalty</p></li>
<li><p><a class="reference internal" href="modules/generated/sklearn.linear_model.Lasso.html#sklearn.linear_model.Lasso" title="sklearn.linear_model.Lasso"><code class="xref py py-class docutils literal notranslate"><span class="pre">Lasso</span></code></a>, L1 penalty (sparse models)</p></li>
<li><p><a class="reference internal" href="modules/generated/sklearn.linear_model.ElasticNet.html#sklearn.linear_model.ElasticNet" title="sklearn.linear_model.ElasticNet"><code class="xref py py-class docutils literal notranslate"><span class="pre">ElasticNet</span></code></a>, L1 + L2 penalty (less sparse models)</p></li>
<li><p><a class="reference internal" href="modules/generated/sklearn.linear_model.SGDRegressor.html#sklearn.linear_model.SGDRegressor" title="sklearn.linear_model.SGDRegressor"><code class="xref py py-class docutils literal notranslate"><span class="pre">SGDRegressor</span></code></a> with <code class="docutils literal notranslate"><span class="pre">loss='squared_loss'</span></code></p></li>
</ul>
<p><strong>Maintainer perspective:</strong>
They all do in principle the same and are different only by the penalty they
impose. This, however, has a large impact on the way the underlying
optimization problem is solved. In the end, this amounts to usage of different
methods and tricks from linear algebra. A special case is <code class="docutils literal notranslate"><span class="pre">SGDRegressor</span></code> which
comprises all 4 previous models and is different by the optimization procedure.
A further side effect is that the different estimators favor different data
layouts (<code class="docutils literal notranslate"><span class="pre">X</span></code> c-contiguous or f-contiguous, sparse csr or csc). This complexity
of the seemingly simple linear models is the reason for having different
estimator classes for different penalties.</p>
<p><strong>User perspective:</strong>
First, the current design is inspired by the scientific literature where linear
regression models with different regularization/penalty were given different
names, e.g. <em>ridge regression</em>. Having different model classes with according
names makes it easier for users to find those regression models.
Secondly, if all the 5 above mentioned linear models were unified into a single
class, there would be parameters with a lot of options like the <code class="docutils literal notranslate"><span class="pre">solver</span></code>
parameter. On top of that, there would be a lot of exclusive interactions
between different parameters. For example, the possible options of the
parameters <code class="docutils literal notranslate"><span class="pre">solver</span></code>, <code class="docutils literal notranslate"><span class="pre">precompute</span></code> and <code class="docutils literal notranslate"><span class="pre">selection</span></code> would depend on the
chosen values of the penalty parameters <code class="docutils literal notranslate"><span class="pre">alpha</span></code> and <code class="docutils literal notranslate"><span class="pre">l1_ratio</span></code>.</p>
</section>
</section>
<section id="contributing">
<h2><a class="toc-backref" href="#id16" role="doc-backlink">Contributing</a><a class="headerlink" href="#contributing" title="Permalink to this heading">¶</a></h2>
<section id="how-can-i-contribute-to-scikit-learn">
<h3><a class="toc-backref" href="#id17" role="doc-backlink">How can I contribute to scikit-learn?</a><a class="headerlink" href="#how-can-i-contribute-to-scikit-learn" title="Permalink to this heading">¶</a></h3>
<p>See <a class="reference internal" href="developers/contributing.html#contributing"><span class="std std-ref">Contributing</span></a>. Before wanting to add a new algorithm, which is
usually a major and lengthy undertaking, it is recommended to start with
<a class="reference internal" href="developers/contributing.html#new-contributors"><span class="std std-ref">known issues</span></a>. Please do not contact the contributors
of scikit-learn directly regarding contributing to scikit-learn.</p>
</section>
<section id="why-is-my-pull-request-not-getting-any-attention">
<h3><a class="toc-backref" href="#id18" role="doc-backlink">Why is my pull request not getting any attention?</a><a class="headerlink" href="#why-is-my-pull-request-not-getting-any-attention" title="Permalink to this heading">¶</a></h3>
<p>The scikit-learn review process takes a significant amount of time, and
contributors should not be discouraged by a lack of activity or review on
their pull request. We care a lot about getting things right
the first time, as maintenance and later change comes at a high cost.
We rarely release any “experimental” code, so all of our contributions
will be subject to high use immediately and should be of the highest
quality possible initially.</p>
<p>Beyond that, scikit-learn is limited in its reviewing bandwidth; many of the
reviewers and core developers are working on scikit-learn on their own time.
If a review of your pull request comes slowly, it is likely because the
reviewers are busy. We ask for your understanding and request that you
not close your pull request or discontinue your work solely because of
this reason.</p>
</section>
<section id="what-are-the-inclusion-criteria-for-new-algorithms">
<span id="new-algorithms-inclusion-criteria"></span><h3><a class="toc-backref" href="#id19" role="doc-backlink">What are the inclusion criteria for new algorithms ?</a><a class="headerlink" href="#what-are-the-inclusion-criteria-for-new-algorithms" title="Permalink to this heading">¶</a></h3>
<p>We only consider well-established algorithms for inclusion. A rule of thumb is
at least 3 years since publication, 200+ citations, and wide use and
usefulness. A technique that provides a clear-cut improvement (e.g. an
enhanced data structure or a more efficient approximation technique) on
a widely-used method will also be considered for inclusion.</p>
<p>From the algorithms or techniques that meet the above criteria, only those
which fit well within the current API of scikit-learn, that is a <code class="docutils literal notranslate"><span class="pre">fit</span></code>,
<code class="docutils literal notranslate"><span class="pre">predict/transform</span></code> interface and ordinarily having input/output that is a
numpy array or sparse matrix, are accepted.</p>
<p>The contributor should support the importance of the proposed addition with
research papers and/or implementations in other similar packages, demonstrate
its usefulness via common use-cases/applications and corroborate performance
improvements, if any, with benchmarks and/or plots. It is expected that the
proposed algorithm should outperform the methods that are already implemented
in scikit-learn at least in some areas.</p>
<p>Inclusion of a new algorithm speeding up an existing model is easier if:</p>
<ul class="simple">
<li><p>it does not introduce new hyper-parameters (as it makes the library
more future-proof),</p></li>
<li><p>it is easy to document clearly when the contribution improves the speed
and when it does not, for instance “when n_features >>
n_samples”,</p></li>
<li><p>benchmarks clearly show a speed up.</p></li>
</ul>
<p>Also, note that your implementation need not be in scikit-learn to be used
together with scikit-learn tools. You can implement your favorite algorithm
in a scikit-learn compatible way, upload it to GitHub and let us know. We
will be happy to list it under <a class="reference internal" href="related_projects.html#related-projects"><span class="std std-ref">Related Projects</span></a>. If you already have
a package on GitHub following the scikit-learn API, you may also be
interested to look at <a class="reference external" href="https://scikit-learn-contrib.github.io">scikit-learn-contrib</a>.</p>
</section>
<section id="why-are-you-so-selective-on-what-algorithms-you-include-in-scikit-learn">
<span id="selectiveness"></span><h3><a class="toc-backref" href="#id20" role="doc-backlink">Why are you so selective on what algorithms you include in scikit-learn?</a><a class="headerlink" href="#why-are-you-so-selective-on-what-algorithms-you-include-in-scikit-learn" title="Permalink to this heading">¶</a></h3>
<p>Code comes with maintenance cost, and we need to balance the amount of
code we have with the size of the team (and add to this the fact that
complexity scales non linearly with the number of features).
The package relies on core developers using their free time to
fix bugs, maintain code and review contributions.
Any algorithm that is added needs future attention by the developers,
at which point the original author might long have lost interest.
See also <a class="reference internal" href="#new-algorithms-inclusion-criteria"><span class="std std-ref">What are the inclusion criteria for new algorithms ?</span></a>. For a great read about
long-term maintenance issues in open-source software, look at
<a class="reference external" href="https://www.fordfoundation.org/media/2976/roads-and-bridges-the-unseen-labor-behind-our-digital-infrastructure.pdf#page=8">the Executive Summary of Roads and Bridges</a></p>
</section>
</section>
<section id="using-scikit-learn">
<h2><a class="toc-backref" href="#id21" role="doc-backlink">Using scikit-learn</a><a class="headerlink" href="#using-scikit-learn" title="Permalink to this heading">¶</a></h2>
<section id="what-s-the-best-way-to-get-help-on-scikit-learn-usage">
<h3><a class="toc-backref" href="#id22" role="doc-backlink">What’s the best way to get help on scikit-learn usage?</a><a class="headerlink" href="#what-s-the-best-way-to-get-help-on-scikit-learn-usage" title="Permalink to this heading">¶</a></h3>
<p><strong>For general machine learning questions</strong>, please use
<a class="reference external" href="https://stats.stackexchange.com/">Cross Validated</a> with the <code class="docutils literal notranslate"><span class="pre">[machine-learning]</span></code> tag.</p>
<p><strong>For scikit-learn usage questions</strong>, please use <a class="reference external" href="https://stackoverflow.com/questions/tagged/scikit-learn">Stack Overflow</a>
with the <code class="docutils literal notranslate"><span class="pre">[scikit-learn]</span></code> and <code class="docutils literal notranslate"><span class="pre">[python]</span></code> tags. You can alternatively use the <a class="reference external" href="https://mail.python.org/mailman/listinfo/scikit-learn">mailing list</a>.</p>
<p>Please make sure to include a minimal reproduction code snippet (ideally shorter
than 10 lines) that highlights your problem on a toy dataset (for instance from
<code class="docutils literal notranslate"><span class="pre">sklearn.datasets</span></code> or randomly generated with functions of <code class="docutils literal notranslate"><span class="pre">numpy.random</span></code> with
a fixed random seed). Please remove any line of code that is not necessary to
reproduce your problem.</p>
<p>The problem should be reproducible by simply copy-pasting your code snippet in a Python
shell with scikit-learn installed. Do not forget to include the import statements.</p>
<p>More guidance to write good reproduction code snippets can be found at:</p>
<p><a class="reference external" href="https://stackoverflow.com/help/mcve">https://stackoverflow.com/help/mcve</a></p>
<p>If your problem raises an exception that you do not understand (even after googling it),
please make sure to include the full traceback that you obtain when running the
reproduction script.</p>
<p>For bug reports or feature requests, please make use of the
<a class="reference external" href="https://github.com/scikit-learn/scikit-learn/issues">issue tracker on GitHub</a>.</p>
<p>There is also a <a class="reference external" href="https://gitter.im/scikit-learn/scikit-learn">scikit-learn Gitter channel</a> where some users and developers
might be found.</p>
<p><strong>Please do not email any authors directly to ask for assistance, report bugs,
or for any other issue related to scikit-learn.</strong></p>
</section>
<section id="how-should-i-save-export-or-deploy-estimators-for-production">
<h3><a class="toc-backref" href="#id23" role="doc-backlink">How should I save, export or deploy estimators for production?</a><a class="headerlink" href="#how-should-i-save-export-or-deploy-estimators-for-production" title="Permalink to this heading">¶</a></h3>
<p>See <a class="reference internal" href="model_persistence.html#model-persistence"><span class="std std-ref">Model persistence</span></a>.</p>
</section>
<section id="how-can-i-create-a-bunch-object">
<h3><a class="toc-backref" href="#id24" role="doc-backlink">How can I create a bunch object?</a><a class="headerlink" href="#how-can-i-create-a-bunch-object" title="Permalink to this heading">¶</a></h3>
<p>Bunch objects are sometimes used as an output for functions and methods. They
extend dictionaries by enabling values to be accessed by key,
<code class="docutils literal notranslate"><span class="pre">bunch["value_key"]</span></code>, or by an attribute, <code class="docutils literal notranslate"><span class="pre">bunch.value_key</span></code>.</p>
<p>They should not be used as an input; therefore you almost never need to create
a <code class="docutils literal notranslate"><span class="pre">Bunch</span></code> object, unless you are extending the scikit-learn’s API.</p>
</section>
<section id="how-can-i-load-my-own-datasets-into-a-format-usable-by-scikit-learn">
<h3><a class="toc-backref" href="#id25" role="doc-backlink">How can I load my own datasets into a format usable by scikit-learn?</a><a class="headerlink" href="#how-can-i-load-my-own-datasets-into-a-format-usable-by-scikit-learn" title="Permalink to this heading">¶</a></h3>
<p>Generally, scikit-learn works on any numeric data stored as numpy arrays
or scipy sparse matrices. Other types that are convertible to numeric
arrays such as pandas DataFrame are also acceptable.</p>
<p>For more information on loading your data files into these usable data
structures, please refer to <a class="reference internal" href="datasets/loading_other_datasets.html#external-datasets"><span class="std std-ref">loading external datasets</span></a>.</p>
</section>
<section id="how-do-i-deal-with-string-data-or-trees-graphs">
<h3><a class="toc-backref" href="#id26" role="doc-backlink">How do I deal with string data (or trees, graphs…)?</a><a class="headerlink" href="#how-do-i-deal-with-string-data-or-trees-graphs" title="Permalink to this heading">¶</a></h3>
<p>scikit-learn estimators assume you’ll feed them real-valued feature vectors.
This assumption is hard-coded in pretty much all of the library.
However, you can feed non-numerical inputs to estimators in several ways.</p>
<p>If you have text documents, you can use a term frequency features; see
<a class="reference internal" href="modules/feature_extraction.html#text-feature-extraction"><span class="std std-ref">Text feature extraction</span></a> for the built-in <em>text vectorizers</em>.
For more general feature extraction from any kind of data, see
<a class="reference internal" href="modules/feature_extraction.html#dict-feature-extraction"><span class="std std-ref">Loading features from dicts</span></a> and <a class="reference internal" href="modules/feature_extraction.html#feature-hashing"><span class="std std-ref">Feature hashing</span></a>.</p>
<p>Another common case is when you have non-numerical data and a custom distance
(or similarity) metric on these data. Examples include strings with edit
distance (aka. Levenshtein distance; e.g., DNA or RNA sequences). These can be
encoded as numbers, but doing so is painful and error-prone. Working with
distance metrics on arbitrary data can be done in two ways.</p>
<p>Firstly, many estimators take precomputed distance/similarity matrices, so if
the dataset is not too large, you can compute distances for all pairs of inputs.
If the dataset is large, you can use feature vectors with only one “feature”,
which is an index into a separate data structure, and supply a custom metric
function that looks up the actual data in this data structure. E.g., to use
DBSCAN with Levenshtein distances:</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="gp">>>> </span><span class="kn">from</span> <span class="nn">leven</span> <span class="kn">import</span> <span class="n">levenshtein</span>
<span class="gp">>>> </span><span class="kn">import</span> <span class="nn">numpy</span> <span class="k">as</span> <span class="nn">np</span>
<span class="gp">>>> </span><span class="kn">from</span> <span class="nn">sklearn.cluster</span> <span class="kn">import</span> <span class="n">dbscan</span>
<span class="gp">>>> </span><span class="n">data</span> <span class="o">=</span> <span class="p">[</span><span class="s2">"ACCTCCTAGAAG"</span><span class="p">,</span> <span class="s2">"ACCTACTAGAAGTT"</span><span class="p">,</span> <span class="s2">"GAATATTAGGCCGA"</span><span class="p">]</span>
<span class="gp">>>> </span><span class="k">def</span> <span class="nf">lev_metric</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">y</span><span class="p">):</span>
<span class="gp">... </span> <span class="n">i</span><span class="p">,</span> <span class="n">j</span> <span class="o">=</span> <span class="nb">int</span><span class="p">(</span><span class="n">x</span><span class="p">[</span><span class="mi">0</span><span class="p">]),</span> <span class="nb">int</span><span class="p">(</span><span class="n">y</span><span class="p">[</span><span class="mi">0</span><span class="p">])</span> <span class="c1"># extract indices</span>
<span class="gp">... </span> <span class="k">return</span> <span class="n">levenshtein</span><span class="p">(</span><span class="n">data</span><span class="p">[</span><span class="n">i</span><span class="p">],</span> <span class="n">data</span><span class="p">[</span><span class="n">j</span><span class="p">])</span>
<span class="gp">...</span>
<span class="gp">>>> </span><span class="n">X</span> <span class="o">=</span> <span class="n">np</span><span class="o">.</span><span class="n">arange</span><span class="p">(</span><span class="nb">len</span><span class="p">(</span><span class="n">data</span><span class="p">))</span><span class="o">.</span><span class="n">reshape</span><span class="p">(</span><span class="o">-</span><span class="mi">1</span><span class="p">,</span> <span class="mi">1</span><span class="p">)</span>
<span class="gp">>>> </span><span class="n">X</span>
<span class="go">array([[0],</span>
<span class="go"> [1],</span>
<span class="go"> [2]])</span>
<span class="gp">>>> </span><span class="c1"># We need to specify algorithm='brute' as the default assumes</span>
<span class="gp">>>> </span><span class="c1"># a continuous feature space.</span>
<span class="gp">>>> </span><span class="n">dbscan</span><span class="p">(</span><span class="n">X</span><span class="p">,</span> <span class="n">metric</span><span class="o">=</span><span class="n">lev_metric</span><span class="p">,</span> <span class="n">eps</span><span class="o">=</span><span class="mi">5</span><span class="p">,</span> <span class="n">min_samples</span><span class="o">=</span><span class="mi">2</span><span class="p">,</span> <span class="n">algorithm</span><span class="o">=</span><span class="s1">'brute'</span><span class="p">)</span>
<span class="gp">... </span>
<span class="go">([0, 1], array([ 0, 0, -1]))</span>
</pre></div>
</div>
<p>(This uses the third-party edit distance package <code class="docutils literal notranslate"><span class="pre">leven</span></code>.)</p>
<p>Similar tricks can be used, with some care, for tree kernels, graph kernels,
etc.</p>
</section>
<section id="why-do-i-sometime-get-a-crash-freeze-with-n-jobs-1-under-osx-or-linux">
<h3><a class="toc-backref" href="#id27" role="doc-backlink">Why do I sometime get a crash/freeze with n_jobs > 1 under OSX or Linux?</a><a class="headerlink" href="#why-do-i-sometime-get-a-crash-freeze-with-n-jobs-1-under-osx-or-linux" title="Permalink to this heading">¶</a></h3>
<p>Several scikit-learn tools such as <code class="docutils literal notranslate"><span class="pre">GridSearchCV</span></code> and <code class="docutils literal notranslate"><span class="pre">cross_val_score</span></code>
rely internally on Python’s <code class="docutils literal notranslate"><span class="pre">multiprocessing</span></code> module to parallelize execution
onto several Python processes by passing <code class="docutils literal notranslate"><span class="pre">n_jobs</span> <span class="pre">></span> <span class="pre">1</span></code> as an argument.</p>
<p>The problem is that Python <code class="docutils literal notranslate"><span class="pre">multiprocessing</span></code> does a <code class="docutils literal notranslate"><span class="pre">fork</span></code> system call
without following it with an <code class="docutils literal notranslate"><span class="pre">exec</span></code> system call for performance reasons. Many
libraries like (some versions of) Accelerate / vecLib under OSX, (some versions
of) MKL, the OpenMP runtime of GCC, nvidia’s Cuda (and probably many others),
manage their own internal thread pool. Upon a call to <code class="docutils literal notranslate"><span class="pre">fork</span></code>, the thread pool
state in the child process is corrupted: the thread pool believes it has many
threads while only the main thread state has been forked. It is possible to
change the libraries to make them detect when a fork happens and reinitialize
the thread pool in that case: we did that for OpenBLAS (merged upstream in
main since 0.2.10) and we contributed a <a class="reference external" href="https://gcc.gnu.org/bugzilla/show_bug.cgi?id=60035">patch</a> to GCC’s OpenMP runtime
(not yet reviewed).</p>
<p>But in the end the real culprit is Python’s <code class="docutils literal notranslate"><span class="pre">multiprocessing</span></code> that does
<code class="docutils literal notranslate"><span class="pre">fork</span></code> without <code class="docutils literal notranslate"><span class="pre">exec</span></code> to reduce the overhead of starting and using new
Python processes for parallel computing. Unfortunately this is a violation of
the POSIX standard and therefore some software editors like Apple refuse to
consider the lack of fork-safety in Accelerate / vecLib as a bug.</p>
<p>In Python 3.4+ it is now possible to configure <code class="docutils literal notranslate"><span class="pre">multiprocessing</span></code> to
use the ‘forkserver’ or ‘spawn’ start methods (instead of the default
‘fork’) to manage the process pools. To work around this issue when
using scikit-learn, you can set the <code class="docutils literal notranslate"><span class="pre">JOBLIB_START_METHOD</span></code> environment
variable to ‘forkserver’. However the user should be aware that using
the ‘forkserver’ method prevents joblib.Parallel to call function
interactively defined in a shell session.</p>
<p>If you have custom code that uses <code class="docutils literal notranslate"><span class="pre">multiprocessing</span></code> directly instead of using
it via joblib you can enable the ‘forkserver’ mode globally for your
program: Insert the following instructions in your main script:</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="kn">import</span> <span class="nn">multiprocessing</span>
<span class="c1"># other imports, custom code, load data, define model...</span>
<span class="k">if</span> <span class="vm">__name__</span> <span class="o">==</span> <span class="s1">'__main__'</span><span class="p">:</span>
<span class="n">multiprocessing</span><span class="o">.</span><span class="n">set_start_method</span><span class="p">(</span><span class="s1">'forkserver'</span><span class="p">)</span>
<span class="c1"># call scikit-learn utils with n_jobs > 1 here</span>
</pre></div>
</div>
<p>You can find more default on the new start methods in the <a class="reference external" href="https://docs.python.org/3/library/multiprocessing.html#contexts-and-start-methods">multiprocessing
documentation</a>.</p>
</section>
<section id="why-does-my-job-use-more-cores-than-specified-with-n-jobs">
<span id="faq-mkl-threading"></span><h3><a class="toc-backref" href="#id28" role="doc-backlink">Why does my job use more cores than specified with n_jobs?</a><a class="headerlink" href="#why-does-my-job-use-more-cores-than-specified-with-n-jobs" title="Permalink to this heading">¶</a></h3>
<p>This is because <code class="docutils literal notranslate"><span class="pre">n_jobs</span></code> only controls the number of jobs for
routines that are parallelized with <code class="docutils literal notranslate"><span class="pre">joblib</span></code>, but parallel code can come
from other sources:</p>
<ul class="simple">
<li><p>some routines may be parallelized with OpenMP (for code written in C or
Cython).</p></li>
<li><p>scikit-learn relies a lot on numpy, which in turn may rely on numerical
libraries like MKL, OpenBLAS or BLIS which can provide parallel
implementations.</p></li>
</ul>
<p>For more details, please refer to our <a class="reference internal" href="computing/parallelism.html#parallelism"><span class="std std-ref">Parallelism notes</span></a>.</p>
</section>
<section id="how-do-i-set-a-random-state-for-an-entire-execution">
<h3><a class="toc-backref" href="#id29" role="doc-backlink">How do I set a <code class="docutils literal notranslate"><span class="pre">random_state</span></code> for an entire execution?</a><a class="headerlink" href="#how-do-i-set-a-random-state-for-an-entire-execution" title="Permalink to this heading">¶</a></h3>
<p>Please refer to <a class="reference internal" href="common_pitfalls.html#randomness"><span class="std std-ref">Controlling randomness</span></a>.</p>
</section>
</section>
</section>
</div>
<div class="container">
<footer class="sk-content-footer">
© 2007 - 2023, scikit-learn developers (BSD License).
<a href="_sources/faq.rst.txt" rel="nofollow">Show this page source</a>
</footer>
</div>
</div>
</div>
<script src="_static/js/vendor/bootstrap.min.js"></script>
<script>
window.ga=window.ga||function(){(ga.q=ga.q||[]).push(arguments)};ga.l=+new Date;
ga('create', 'UA-22606712-2', 'auto');
ga('set', 'anonymizeIp', true);
ga('send', 'pageview');
</script>
<script async src='https://www.google-analytics.com/analytics.js'></script>
<script defer data-domain="scikit-learn.org" src="https://views.scientific-python.org/js/script.js">
</script>
<script src="_static/clipboard.min.js"></script>
<script src="_static/copybutton.js"></script>
<script>
$(document).ready(function() {
/* Add a [>>>] button on the top-right corner of code samples to hide
* the >>> and ... prompts and the output and thus make the code
* copyable. */
var div = $('.highlight-python .highlight,' +
'.highlight-python3 .highlight,' +
'.highlight-pycon .highlight,' +
'.highlight-default .highlight')
var pre = div.find('pre');
// get the styles from the current theme
pre.parent().parent().css('position', 'relative');
// create and add the button to all the code blocks that contain >>>
div.each(function(index) {
var jthis = $(this);
// tracebacks (.gt) contain bare text elements that need to be
// wrapped in a span to work with .nextUntil() (see later)
jthis.find('pre:has(.gt)').contents().filter(function() {
return ((this.nodeType == 3) && (this.data.trim().length > 0));
}).wrap('<span>');
});
/*** Add permalink buttons next to glossary terms ***/
$('dl.glossary > dt[id]').append(function() {
return ('<a class="headerlink" href="#' +
this.getAttribute('id') +
'" title="Permalink to this term">¶</a>');
});
});
</script>
<script id="MathJax-script" async src="https://cdn.jsdelivr.net/npm/mathjax@3/es5/tex-chtml.js"></script>
<script src="https://scikit-learn.org/versionwarning.js"></script>
</body>
</html>