forked from scikit-learn/scikit-learn.github.io
-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathimpute.html
637 lines (605 loc) · 62.8 KB
/
impute.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
<!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="6.4. Imputation of missing values" />
<meta property="og:type" content="website" />
<meta property="og:url" content="https://scikit-learn/stable/modules/impute.html" />
<meta property="og:site_name" content="scikit-learn" />
<meta property="og:description" content="For various reasons, many real world datasets contain missing values, often encoded as blanks, NaNs or other placeholders. Such datasets however are incompatible with scikit-learn estimators which ..." />
<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="For various reasons, many real world datasets contain missing values, often encoded as blanks, NaNs or other placeholders. Such datasets however are incompatible with scikit-learn estimators which ..." />
<meta name="viewport" content="width=device-width, initial-scale=1.0">
<title>6.4. Imputation of missing values — scikit-learn 1.4.dev0 documentation</title>
<link rel="canonical" href="http://scikit-learn.org/stable/modules/impute.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="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.html" >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.html" >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="preprocessing.html" role="button" class="btn sk-btn-rellink py-1" sk-rellink-tooltip="6.3. Preprocessing data">Prev</a><a href="../data_transforms.html" role="button" class="btn sk-btn-rellink py-1" sk-rellink-tooltip="6. Dataset transformations">Up</a>
<a href="unsupervised_reduction.html" role="button" class="btn sk-btn-rellink py-1" sk-rellink-tooltip="6.5. Unsupervised dimensionality reduction">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="#">6.4. Imputation of missing values</a><ul>
<li><a class="reference internal" href="#univariate-vs-multivariate-imputation">6.4.1. Univariate vs. Multivariate Imputation</a></li>
<li><a class="reference internal" href="#univariate-feature-imputation">6.4.2. Univariate feature imputation</a></li>
<li><a class="reference internal" href="#multivariate-feature-imputation">6.4.3. Multivariate feature imputation</a><ul>
<li><a class="reference internal" href="#flexibility-of-iterativeimputer">6.4.3.1. Flexibility of IterativeImputer</a></li>
<li><a class="reference internal" href="#multiple-vs-single-imputation">6.4.3.2. Multiple vs. Single Imputation</a></li>
<li><a class="reference internal" href="#references">6.4.3.3. References</a></li>
</ul>
</li>
<li><a class="reference internal" href="#nearest-neighbors-imputation">6.4.4. Nearest neighbors imputation</a></li>
<li><a class="reference internal" href="#keeping-the-number-of-features-constant">6.4.5. Keeping the number of features constant</a></li>
<li><a class="reference internal" href="#marking-imputed-values">6.4.6. Marking imputed values</a></li>
<li><a class="reference internal" href="#estimators-that-handle-nan-values">6.4.7. Estimators that handle NaN values</a></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="imputation-of-missing-values">
<span id="impute"></span><h1><span class="section-number">6.4. </span>Imputation of missing values<a class="headerlink" href="#imputation-of-missing-values" title="Permalink to this heading">¶</a></h1>
<p>For various reasons, many real world datasets contain missing values, often
encoded as blanks, NaNs or other placeholders. Such datasets however are
incompatible with scikit-learn estimators which assume that all values in an
array are numerical, and that all have and hold meaning. A basic strategy to
use incomplete datasets is to discard entire rows and/or columns containing
missing values. However, this comes at the price of losing data which may be
valuable (even though incomplete). A better strategy is to impute the missing
values, i.e., to infer them from the known part of the data. See the
glossary entry on <a class="reference internal" href="../glossary.html#term-imputation"><span class="xref std std-term">imputation</span></a>.</p>
<section id="univariate-vs-multivariate-imputation">
<h2><span class="section-number">6.4.1. </span>Univariate vs. Multivariate Imputation<a class="headerlink" href="#univariate-vs-multivariate-imputation" title="Permalink to this heading">¶</a></h2>
<p>One type of imputation algorithm is univariate, which imputes values in the
i-th feature dimension using only non-missing values in that feature dimension
(e.g. <a class="reference internal" href="generated/sklearn.impute.SimpleImputer.html#sklearn.impute.SimpleImputer" title="sklearn.impute.SimpleImputer"><code class="xref py py-class docutils literal notranslate"><span class="pre">SimpleImputer</span></code></a>). By contrast, multivariate imputation
algorithms use the entire set of available feature dimensions to estimate the
missing values (e.g. <a class="reference internal" href="generated/sklearn.impute.IterativeImputer.html#sklearn.impute.IterativeImputer" title="sklearn.impute.IterativeImputer"><code class="xref py py-class docutils literal notranslate"><span class="pre">IterativeImputer</span></code></a>).</p>
</section>
<section id="univariate-feature-imputation">
<span id="single-imputer"></span><h2><span class="section-number">6.4.2. </span>Univariate feature imputation<a class="headerlink" href="#univariate-feature-imputation" title="Permalink to this heading">¶</a></h2>
<p>The <a class="reference internal" href="generated/sklearn.impute.SimpleImputer.html#sklearn.impute.SimpleImputer" title="sklearn.impute.SimpleImputer"><code class="xref py py-class docutils literal notranslate"><span class="pre">SimpleImputer</span></code></a> class provides basic strategies for imputing missing
values. Missing values can be imputed with a provided constant value, or using
the statistics (mean, median or most frequent) of each column in which the
missing values are located. This class also allows for different missing values
encodings.</p>
<p>The following snippet demonstrates how to replace missing values,
encoded as <code class="docutils literal notranslate"><span class="pre">np.nan</span></code>, using the mean value of the columns (axis 0)
that contain the missing values:</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></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.impute</span> <span class="kn">import</span> <span class="n">SimpleImputer</span>
<span class="gp">>>> </span><span class="n">imp</span> <span class="o">=</span> <span class="n">SimpleImputer</span><span class="p">(</span><span class="n">missing_values</span><span class="o">=</span><span class="n">np</span><span class="o">.</span><span class="n">nan</span><span class="p">,</span> <span class="n">strategy</span><span class="o">=</span><span class="s1">'mean'</span><span class="p">)</span>
<span class="gp">>>> </span><span class="n">imp</span><span class="o">.</span><span class="n">fit</span><span class="p">([[</span><span class="mi">1</span><span class="p">,</span> <span class="mi">2</span><span class="p">],</span> <span class="p">[</span><span class="n">np</span><span class="o">.</span><span class="n">nan</span><span class="p">,</span> <span class="mi">3</span><span class="p">],</span> <span class="p">[</span><span class="mi">7</span><span class="p">,</span> <span class="mi">6</span><span class="p">]])</span>
<span class="go">SimpleImputer()</span>
<span class="gp">>>> </span><span class="n">X</span> <span class="o">=</span> <span class="p">[[</span><span class="n">np</span><span class="o">.</span><span class="n">nan</span><span class="p">,</span> <span class="mi">2</span><span class="p">],</span> <span class="p">[</span><span class="mi">6</span><span class="p">,</span> <span class="n">np</span><span class="o">.</span><span class="n">nan</span><span class="p">],</span> <span class="p">[</span><span class="mi">7</span><span class="p">,</span> <span class="mi">6</span><span class="p">]]</span>
<span class="gp">>>> </span><span class="nb">print</span><span class="p">(</span><span class="n">imp</span><span class="o">.</span><span class="n">transform</span><span class="p">(</span><span class="n">X</span><span class="p">))</span>
<span class="go">[[4. 2. ]</span>
<span class="go"> [6. 3.666...]</span>
<span class="go"> [7. 6. ]]</span>
</pre></div>
</div>
<p>The <a class="reference internal" href="generated/sklearn.impute.SimpleImputer.html#sklearn.impute.SimpleImputer" title="sklearn.impute.SimpleImputer"><code class="xref py py-class docutils literal notranslate"><span class="pre">SimpleImputer</span></code></a> class also supports sparse matrices:</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="gp">>>> </span><span class="kn">import</span> <span class="nn">scipy.sparse</span> <span class="k">as</span> <span class="nn">sp</span>
<span class="gp">>>> </span><span class="n">X</span> <span class="o">=</span> <span class="n">sp</span><span class="o">.</span><span class="n">csc_matrix</span><span class="p">([[</span><span class="mi">1</span><span class="p">,</span> <span class="mi">2</span><span class="p">],</span> <span class="p">[</span><span class="mi">0</span><span class="p">,</span> <span class="o">-</span><span class="mi">1</span><span class="p">],</span> <span class="p">[</span><span class="mi">8</span><span class="p">,</span> <span class="mi">4</span><span class="p">]])</span>
<span class="gp">>>> </span><span class="n">imp</span> <span class="o">=</span> <span class="n">SimpleImputer</span><span class="p">(</span><span class="n">missing_values</span><span class="o">=-</span><span class="mi">1</span><span class="p">,</span> <span class="n">strategy</span><span class="o">=</span><span class="s1">'mean'</span><span class="p">)</span>
<span class="gp">>>> </span><span class="n">imp</span><span class="o">.</span><span class="n">fit</span><span class="p">(</span><span class="n">X</span><span class="p">)</span>
<span class="go">SimpleImputer(missing_values=-1)</span>
<span class="gp">>>> </span><span class="n">X_test</span> <span class="o">=</span> <span class="n">sp</span><span class="o">.</span><span class="n">csc_matrix</span><span class="p">([[</span><span class="o">-</span><span class="mi">1</span><span class="p">,</span> <span class="mi">2</span><span class="p">],</span> <span class="p">[</span><span class="mi">6</span><span class="p">,</span> <span class="o">-</span><span class="mi">1</span><span class="p">],</span> <span class="p">[</span><span class="mi">7</span><span class="p">,</span> <span class="mi">6</span><span class="p">]])</span>
<span class="gp">>>> </span><span class="nb">print</span><span class="p">(</span><span class="n">imp</span><span class="o">.</span><span class="n">transform</span><span class="p">(</span><span class="n">X_test</span><span class="p">)</span><span class="o">.</span><span class="n">toarray</span><span class="p">())</span>
<span class="go">[[3. 2.]</span>
<span class="go"> [6. 3.]</span>
<span class="go"> [7. 6.]]</span>
</pre></div>
</div>
<p>Note that this format is not meant to be used to implicitly store missing
values in the matrix because it would densify it at transform time. Missing
values encoded by 0 must be used with dense input.</p>
<p>The <a class="reference internal" href="generated/sklearn.impute.SimpleImputer.html#sklearn.impute.SimpleImputer" title="sklearn.impute.SimpleImputer"><code class="xref py py-class docutils literal notranslate"><span class="pre">SimpleImputer</span></code></a> class also supports categorical data represented as
string values or pandas categoricals when using the <code class="docutils literal notranslate"><span class="pre">'most_frequent'</span></code> or
<code class="docutils literal notranslate"><span class="pre">'constant'</span></code> strategy:</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="gp">>>> </span><span class="kn">import</span> <span class="nn">pandas</span> <span class="k">as</span> <span class="nn">pd</span>
<span class="gp">>>> </span><span class="n">df</span> <span class="o">=</span> <span class="n">pd</span><span class="o">.</span><span class="n">DataFrame</span><span class="p">([[</span><span class="s2">"a"</span><span class="p">,</span> <span class="s2">"x"</span><span class="p">],</span>
<span class="gp">... </span> <span class="p">[</span><span class="n">np</span><span class="o">.</span><span class="n">nan</span><span class="p">,</span> <span class="s2">"y"</span><span class="p">],</span>
<span class="gp">... </span> <span class="p">[</span><span class="s2">"a"</span><span class="p">,</span> <span class="n">np</span><span class="o">.</span><span class="n">nan</span><span class="p">],</span>
<span class="gp">... </span> <span class="p">[</span><span class="s2">"b"</span><span class="p">,</span> <span class="s2">"y"</span><span class="p">]],</span> <span class="n">dtype</span><span class="o">=</span><span class="s2">"category"</span><span class="p">)</span>
<span class="gp">...</span>
<span class="gp">>>> </span><span class="n">imp</span> <span class="o">=</span> <span class="n">SimpleImputer</span><span class="p">(</span><span class="n">strategy</span><span class="o">=</span><span class="s2">"most_frequent"</span><span class="p">)</span>
<span class="gp">>>> </span><span class="nb">print</span><span class="p">(</span><span class="n">imp</span><span class="o">.</span><span class="n">fit_transform</span><span class="p">(</span><span class="n">df</span><span class="p">))</span>
<span class="go">[['a' 'x']</span>
<span class="go"> ['a' 'y']</span>
<span class="go"> ['a' 'y']</span>
<span class="go"> ['b' 'y']]</span>
</pre></div>
</div>
</section>
<section id="multivariate-feature-imputation">
<span id="iterative-imputer"></span><h2><span class="section-number">6.4.3. </span>Multivariate feature imputation<a class="headerlink" href="#multivariate-feature-imputation" title="Permalink to this heading">¶</a></h2>
<p>A more sophisticated approach is to use the <a class="reference internal" href="generated/sklearn.impute.IterativeImputer.html#sklearn.impute.IterativeImputer" title="sklearn.impute.IterativeImputer"><code class="xref py py-class docutils literal notranslate"><span class="pre">IterativeImputer</span></code></a> class,
which models each feature with missing values as a function of other features,
and uses that estimate for imputation. It does so in an iterated round-robin
fashion: at each step, a feature column is designated as output <code class="docutils literal notranslate"><span class="pre">y</span></code> and the
other feature columns are treated as inputs <code class="docutils literal notranslate"><span class="pre">X</span></code>. A regressor is fit on <code class="docutils literal notranslate"><span class="pre">(X,</span>
<span class="pre">y)</span></code> for known <code class="docutils literal notranslate"><span class="pre">y</span></code>. Then, the regressor is used to predict the missing values
of <code class="docutils literal notranslate"><span class="pre">y</span></code>. This is done for each feature in an iterative fashion, and then is
repeated for <code class="docutils literal notranslate"><span class="pre">max_iter</span></code> imputation rounds. The results of the final
imputation round are returned.</p>
<div class="admonition note">
<p class="admonition-title">Note</p>
<p>This estimator is still <strong>experimental</strong> for now: default parameters or
details of behaviour might change without any deprecation cycle. Resolving
the following issues would help stabilize <a class="reference internal" href="generated/sklearn.impute.IterativeImputer.html#sklearn.impute.IterativeImputer" title="sklearn.impute.IterativeImputer"><code class="xref py py-class docutils literal notranslate"><span class="pre">IterativeImputer</span></code></a>:
convergence criteria (<a class="reference external" href="https://github.com/scikit-learn/scikit-learn/issues/14338">#14338</a>), default estimators (<a class="reference external" href="https://github.com/scikit-learn/scikit-learn/issues/13286">#13286</a>),
and use of random state (<a class="reference external" href="https://github.com/scikit-learn/scikit-learn/issues/15611">#15611</a>). To use it, you need to explicitly
import <code class="docutils literal notranslate"><span class="pre">enable_iterative_imputer</span></code>.</p>
</div>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></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.experimental</span> <span class="kn">import</span> <span class="n">enable_iterative_imputer</span>
<span class="gp">>>> </span><span class="kn">from</span> <span class="nn">sklearn.impute</span> <span class="kn">import</span> <span class="n">IterativeImputer</span>
<span class="gp">>>> </span><span class="n">imp</span> <span class="o">=</span> <span class="n">IterativeImputer</span><span class="p">(</span><span class="n">max_iter</span><span class="o">=</span><span class="mi">10</span><span class="p">,</span> <span class="n">random_state</span><span class="o">=</span><span class="mi">0</span><span class="p">)</span>
<span class="gp">>>> </span><span class="n">imp</span><span class="o">.</span><span class="n">fit</span><span class="p">([[</span><span class="mi">1</span><span class="p">,</span> <span class="mi">2</span><span class="p">],</span> <span class="p">[</span><span class="mi">3</span><span class="p">,</span> <span class="mi">6</span><span class="p">],</span> <span class="p">[</span><span class="mi">4</span><span class="p">,</span> <span class="mi">8</span><span class="p">],</span> <span class="p">[</span><span class="n">np</span><span class="o">.</span><span class="n">nan</span><span class="p">,</span> <span class="mi">3</span><span class="p">],</span> <span class="p">[</span><span class="mi">7</span><span class="p">,</span> <span class="n">np</span><span class="o">.</span><span class="n">nan</span><span class="p">]])</span>
<span class="go">IterativeImputer(random_state=0)</span>
<span class="gp">>>> </span><span class="n">X_test</span> <span class="o">=</span> <span class="p">[[</span><span class="n">np</span><span class="o">.</span><span class="n">nan</span><span class="p">,</span> <span class="mi">2</span><span class="p">],</span> <span class="p">[</span><span class="mi">6</span><span class="p">,</span> <span class="n">np</span><span class="o">.</span><span class="n">nan</span><span class="p">],</span> <span class="p">[</span><span class="n">np</span><span class="o">.</span><span class="n">nan</span><span class="p">,</span> <span class="mi">6</span><span class="p">]]</span>
<span class="gp">>>> </span><span class="c1"># the model learns that the second feature is double the first</span>
<span class="gp">>>> </span><span class="nb">print</span><span class="p">(</span><span class="n">np</span><span class="o">.</span><span class="n">round</span><span class="p">(</span><span class="n">imp</span><span class="o">.</span><span class="n">transform</span><span class="p">(</span><span class="n">X_test</span><span class="p">)))</span>
<span class="go">[[ 1. 2.]</span>
<span class="go"> [ 6. 12.]</span>
<span class="go"> [ 3. 6.]]</span>
</pre></div>
</div>
<p>Both <a class="reference internal" href="generated/sklearn.impute.SimpleImputer.html#sklearn.impute.SimpleImputer" title="sklearn.impute.SimpleImputer"><code class="xref py py-class docutils literal notranslate"><span class="pre">SimpleImputer</span></code></a> and <a class="reference internal" href="generated/sklearn.impute.IterativeImputer.html#sklearn.impute.IterativeImputer" title="sklearn.impute.IterativeImputer"><code class="xref py py-class docutils literal notranslate"><span class="pre">IterativeImputer</span></code></a> can be used in a
Pipeline as a way to build a composite estimator that supports imputation.
See <a class="reference internal" href="../auto_examples/impute/plot_missing_values.html#sphx-glr-auto-examples-impute-plot-missing-values-py"><span class="std std-ref">Imputing missing values before building an estimator</span></a>.</p>
<section id="flexibility-of-iterativeimputer">
<h3><span class="section-number">6.4.3.1. </span>Flexibility of IterativeImputer<a class="headerlink" href="#flexibility-of-iterativeimputer" title="Permalink to this heading">¶</a></h3>
<p>There are many well-established imputation packages in the R data science
ecosystem: Amelia, mi, mice, missForest, etc. missForest is popular, and turns
out to be a particular instance of different sequential imputation algorithms
that can all be implemented with <a class="reference internal" href="generated/sklearn.impute.IterativeImputer.html#sklearn.impute.IterativeImputer" title="sklearn.impute.IterativeImputer"><code class="xref py py-class docutils literal notranslate"><span class="pre">IterativeImputer</span></code></a> by passing in
different regressors to be used for predicting missing feature values. In the
case of missForest, this regressor is a Random Forest.
See <a class="reference internal" href="../auto_examples/impute/plot_iterative_imputer_variants_comparison.html#sphx-glr-auto-examples-impute-plot-iterative-imputer-variants-comparison-py"><span class="std std-ref">Imputing missing values with variants of IterativeImputer</span></a>.</p>
</section>
<section id="multiple-vs-single-imputation">
<span id="multiple-imputation"></span><h3><span class="section-number">6.4.3.2. </span>Multiple vs. Single Imputation<a class="headerlink" href="#multiple-vs-single-imputation" title="Permalink to this heading">¶</a></h3>
<p>In the statistics community, it is common practice to perform multiple
imputations, generating, for example, <code class="docutils literal notranslate"><span class="pre">m</span></code> separate imputations for a single
feature matrix. Each of these <code class="docutils literal notranslate"><span class="pre">m</span></code> imputations is then put through the
subsequent analysis pipeline (e.g. feature engineering, clustering, regression,
classification). The <code class="docutils literal notranslate"><span class="pre">m</span></code> final analysis results (e.g. held-out validation
errors) allow the data scientist to obtain understanding of how analytic
results may differ as a consequence of the inherent uncertainty caused by the
missing values. The above practice is called multiple imputation.</p>
<p>Our implementation of <a class="reference internal" href="generated/sklearn.impute.IterativeImputer.html#sklearn.impute.IterativeImputer" title="sklearn.impute.IterativeImputer"><code class="xref py py-class docutils literal notranslate"><span class="pre">IterativeImputer</span></code></a> was inspired by the R MICE
package (Multivariate Imputation by Chained Equations) <a class="footnote-reference brackets" href="#id3" id="id1" role="doc-noteref"><span class="fn-bracket">[</span>1<span class="fn-bracket">]</span></a>, but differs from
it by returning a single imputation instead of multiple imputations. However,
<a class="reference internal" href="generated/sklearn.impute.IterativeImputer.html#sklearn.impute.IterativeImputer" title="sklearn.impute.IterativeImputer"><code class="xref py py-class docutils literal notranslate"><span class="pre">IterativeImputer</span></code></a> can also be used for multiple imputations by applying
it repeatedly to the same dataset with different random seeds when
<code class="docutils literal notranslate"><span class="pre">sample_posterior=True</span></code>. See <a class="footnote-reference brackets" href="#id4" id="id2" role="doc-noteref"><span class="fn-bracket">[</span>2<span class="fn-bracket">]</span></a>, chapter 4 for more discussion on multiple
vs. single imputations.</p>
<p>It is still an open problem as to how useful single vs. multiple imputation is
in the context of prediction and classification when the user is not
interested in measuring uncertainty due to missing values.</p>
<p>Note that a call to the <code class="docutils literal notranslate"><span class="pre">transform</span></code> method of <a class="reference internal" href="generated/sklearn.impute.IterativeImputer.html#sklearn.impute.IterativeImputer" title="sklearn.impute.IterativeImputer"><code class="xref py py-class docutils literal notranslate"><span class="pre">IterativeImputer</span></code></a> is
not allowed to change the number of samples. Therefore multiple imputations
cannot be achieved by a single call to <code class="docutils literal notranslate"><span class="pre">transform</span></code>.</p>
</section>
<section id="references">
<h3><span class="section-number">6.4.3.3. </span>References<a class="headerlink" href="#references" title="Permalink to this heading">¶</a></h3>
<aside class="footnote-list brackets">
<aside class="footnote brackets" id="id3" role="note">
<span class="label"><span class="fn-bracket">[</span><a role="doc-backlink" href="#id1">1</a><span class="fn-bracket">]</span></span>
<p>Stef van Buuren, Karin Groothuis-Oudshoorn (2011). “mice: Multivariate
Imputation by Chained Equations in R”. Journal of Statistical Software 45:
1-67.</p>
</aside>
<aside class="footnote brackets" id="id4" role="note">
<span class="label"><span class="fn-bracket">[</span><a role="doc-backlink" href="#id2">2</a><span class="fn-bracket">]</span></span>
<p>Roderick J A Little and Donald B Rubin (1986). “Statistical Analysis
with Missing Data”. John Wiley & Sons, Inc., New York, NY, USA.</p>
</aside>
</aside>
</section>
</section>
<section id="nearest-neighbors-imputation">
<span id="knnimpute"></span><h2><span class="section-number">6.4.4. </span>Nearest neighbors imputation<a class="headerlink" href="#nearest-neighbors-imputation" title="Permalink to this heading">¶</a></h2>
<p>The <a class="reference internal" href="generated/sklearn.impute.KNNImputer.html#sklearn.impute.KNNImputer" title="sklearn.impute.KNNImputer"><code class="xref py py-class docutils literal notranslate"><span class="pre">KNNImputer</span></code></a> class provides imputation for filling in missing values
using the k-Nearest Neighbors approach. By default, a euclidean distance metric
that supports missing values,
<a class="reference internal" href="generated/sklearn.metrics.pairwise.nan_euclidean_distances.html#sklearn.metrics.pairwise.nan_euclidean_distances" title="sklearn.metrics.pairwise.nan_euclidean_distances"><code class="xref py py-func docutils literal notranslate"><span class="pre">nan_euclidean_distances</span></code></a>, is used to find the
nearest neighbors. Each missing feature is imputed using values from
<code class="docutils literal notranslate"><span class="pre">n_neighbors</span></code> nearest neighbors that have a value for the feature. The
feature of the neighbors are averaged uniformly or weighted by distance to each
neighbor. If a sample has more than one feature missing, then the neighbors for
that sample can be different depending on the particular feature being imputed.
When the number of available neighbors is less than <code class="docutils literal notranslate"><span class="pre">n_neighbors</span></code> and there are
no defined distances to the training set, the training set average for that
feature is used during imputation. If there is at least one neighbor with a
defined distance, the weighted or unweighted average of the remaining neighbors
will be used during imputation. If a feature is always missing in training, it
is removed during <code class="docutils literal notranslate"><span class="pre">transform</span></code>. For more information on the methodology, see
ref. <a class="reference internal" href="#ol2001" id="id5"><span>[OL2001]</span></a>.</p>
<p>The following snippet demonstrates how to replace missing values,
encoded as <code class="docutils literal notranslate"><span class="pre">np.nan</span></code>, using the mean feature value of the two nearest
neighbors of samples with missing values:</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></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.impute</span> <span class="kn">import</span> <span class="n">KNNImputer</span>
<span class="gp">>>> </span><span class="n">nan</span> <span class="o">=</span> <span class="n">np</span><span class="o">.</span><span class="n">nan</span>
<span class="gp">>>> </span><span class="n">X</span> <span class="o">=</span> <span class="p">[[</span><span class="mi">1</span><span class="p">,</span> <span class="mi">2</span><span class="p">,</span> <span class="n">nan</span><span class="p">],</span> <span class="p">[</span><span class="mi">3</span><span class="p">,</span> <span class="mi">4</span><span class="p">,</span> <span class="mi">3</span><span class="p">],</span> <span class="p">[</span><span class="n">nan</span><span class="p">,</span> <span class="mi">6</span><span class="p">,</span> <span class="mi">5</span><span class="p">],</span> <span class="p">[</span><span class="mi">8</span><span class="p">,</span> <span class="mi">8</span><span class="p">,</span> <span class="mi">7</span><span class="p">]]</span>
<span class="gp">>>> </span><span class="n">imputer</span> <span class="o">=</span> <span class="n">KNNImputer</span><span class="p">(</span><span class="n">n_neighbors</span><span class="o">=</span><span class="mi">2</span><span class="p">,</span> <span class="n">weights</span><span class="o">=</span><span class="s2">"uniform"</span><span class="p">)</span>
<span class="gp">>>> </span><span class="n">imputer</span><span class="o">.</span><span class="n">fit_transform</span><span class="p">(</span><span class="n">X</span><span class="p">)</span>
<span class="go">array([[1. , 2. , 4. ],</span>
<span class="go"> [3. , 4. , 3. ],</span>
<span class="go"> [5.5, 6. , 5. ],</span>
<span class="go"> [8. , 8. , 7. ]])</span>
</pre></div>
</div>
<aside class="topic">
<p class="topic-title">References</p>
<div role="list" class="citation-list">
<div class="citation" id="ol2001" role="doc-biblioentry">
<span class="label"><span class="fn-bracket">[</span><a role="doc-backlink" href="#id5">OL2001</a><span class="fn-bracket">]</span></span>
<p>Olga Troyanskaya, Michael Cantor, Gavin Sherlock, Pat Brown,
Trevor Hastie, Robert Tibshirani, David Botstein and Russ B. Altman,
Missing value estimation methods for DNA microarrays, BIOINFORMATICS
Vol. 17 no. 6, 2001 Pages 520-525.</p>
</div>
</div>
</aside>
</section>
<section id="keeping-the-number-of-features-constant">
<h2><span class="section-number">6.4.5. </span>Keeping the number of features constant<a class="headerlink" href="#keeping-the-number-of-features-constant" title="Permalink to this heading">¶</a></h2>
<p>By default, the scikit-learn imputers will drop fully empty features, i.e.
columns containing only missing values. For instance:</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="gp">>>> </span><span class="n">imputer</span> <span class="o">=</span> <span class="n">SimpleImputer</span><span class="p">()</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">array</span><span class="p">([[</span><span class="n">np</span><span class="o">.</span><span class="n">nan</span><span class="p">,</span> <span class="mi">1</span><span class="p">],</span> <span class="p">[</span><span class="n">np</span><span class="o">.</span><span class="n">nan</span><span class="p">,</span> <span class="mi">2</span><span class="p">],</span> <span class="p">[</span><span class="n">np</span><span class="o">.</span><span class="n">nan</span><span class="p">,</span> <span class="mi">3</span><span class="p">]])</span>
<span class="gp">>>> </span><span class="n">imputer</span><span class="o">.</span><span class="n">fit_transform</span><span class="p">(</span><span class="n">X</span><span class="p">)</span>
<span class="go">array([[1.],</span>
<span class="go"> [2.],</span>
<span class="go"> [3.]])</span>
</pre></div>
</div>
<p>The first feature in <code class="docutils literal notranslate"><span class="pre">X</span></code> containing only <code class="docutils literal notranslate"><span class="pre">np.nan</span></code> was dropped after the
imputation. While this feature will not help in predictive setting, dropping
the columns will change the shape of <code class="docutils literal notranslate"><span class="pre">X</span></code> which could be problematic when using
imputers in a more complex machine-learning pipeline. The parameter
<code class="docutils literal notranslate"><span class="pre">keep_empty_features</span></code> offers the option to keep the empty features by imputing
with a constant values. In most of the cases, this constant value is zero:</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="gp">>>> </span><span class="n">imputer</span><span class="o">.</span><span class="n">set_params</span><span class="p">(</span><span class="n">keep_empty_features</span><span class="o">=</span><span class="kc">True</span><span class="p">)</span>
<span class="go">SimpleImputer(keep_empty_features=True)</span>
<span class="gp">>>> </span><span class="n">imputer</span><span class="o">.</span><span class="n">fit_transform</span><span class="p">(</span><span class="n">X</span><span class="p">)</span>
<span class="go">array([[0., 1.],</span>
<span class="go"> [0., 2.],</span>
<span class="go"> [0., 3.]])</span>
</pre></div>
</div>
</section>
<section id="marking-imputed-values">
<span id="missing-indicator"></span><h2><span class="section-number">6.4.6. </span>Marking imputed values<a class="headerlink" href="#marking-imputed-values" title="Permalink to this heading">¶</a></h2>
<p>The <a class="reference internal" href="generated/sklearn.impute.MissingIndicator.html#sklearn.impute.MissingIndicator" title="sklearn.impute.MissingIndicator"><code class="xref py py-class docutils literal notranslate"><span class="pre">MissingIndicator</span></code></a> transformer is useful to transform a dataset into
corresponding binary matrix indicating the presence of missing values in the
dataset. This transformation is useful in conjunction with imputation. When
using imputation, preserving the information about which values had been
missing can be informative. Note that both the <a class="reference internal" href="generated/sklearn.impute.SimpleImputer.html#sklearn.impute.SimpleImputer" title="sklearn.impute.SimpleImputer"><code class="xref py py-class docutils literal notranslate"><span class="pre">SimpleImputer</span></code></a> and
<a class="reference internal" href="generated/sklearn.impute.IterativeImputer.html#sklearn.impute.IterativeImputer" title="sklearn.impute.IterativeImputer"><code class="xref py py-class docutils literal notranslate"><span class="pre">IterativeImputer</span></code></a> have the boolean parameter <code class="docutils literal notranslate"><span class="pre">add_indicator</span></code>
(<code class="docutils literal notranslate"><span class="pre">False</span></code> by default) which when set to <code class="docutils literal notranslate"><span class="pre">True</span></code> provides a convenient way of
stacking the output of the <a class="reference internal" href="generated/sklearn.impute.MissingIndicator.html#sklearn.impute.MissingIndicator" title="sklearn.impute.MissingIndicator"><code class="xref py py-class docutils literal notranslate"><span class="pre">MissingIndicator</span></code></a> transformer with the
output of the imputer.</p>
<p><code class="docutils literal notranslate"><span class="pre">NaN</span></code> is usually used as the placeholder for missing values. However, it
enforces the data type to be float. The parameter <code class="docutils literal notranslate"><span class="pre">missing_values</span></code> allows to
specify other placeholder such as integer. In the following example, we will
use <code class="docutils literal notranslate"><span class="pre">-1</span></code> as missing values:</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="gp">>>> </span><span class="kn">from</span> <span class="nn">sklearn.impute</span> <span class="kn">import</span> <span class="n">MissingIndicator</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">array</span><span class="p">([[</span><span class="o">-</span><span class="mi">1</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="mi">3</span><span class="p">],</span>
<span class="gp">... </span> <span class="p">[</span><span class="mi">4</span><span class="p">,</span> <span class="o">-</span><span class="mi">1</span><span class="p">,</span> <span class="mi">0</span><span class="p">,</span> <span class="o">-</span><span class="mi">1</span><span class="p">],</span>
<span class="gp">... </span> <span class="p">[</span><span class="mi">8</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="mi">0</span><span class="p">]])</span>
<span class="gp">>>> </span><span class="n">indicator</span> <span class="o">=</span> <span class="n">MissingIndicator</span><span class="p">(</span><span class="n">missing_values</span><span class="o">=-</span><span class="mi">1</span><span class="p">)</span>
<span class="gp">>>> </span><span class="n">mask_missing_values_only</span> <span class="o">=</span> <span class="n">indicator</span><span class="o">.</span><span class="n">fit_transform</span><span class="p">(</span><span class="n">X</span><span class="p">)</span>
<span class="gp">>>> </span><span class="n">mask_missing_values_only</span>
<span class="go">array([[ True, True, False],</span>
<span class="go"> [False, True, True],</span>
<span class="go"> [False, True, False]])</span>
</pre></div>
</div>
<p>The <code class="docutils literal notranslate"><span class="pre">features</span></code> parameter is used to choose the features for which the mask is
constructed. By default, it is <code class="docutils literal notranslate"><span class="pre">'missing-only'</span></code> which returns the imputer
mask of the features containing missing values at <code class="docutils literal notranslate"><span class="pre">fit</span></code> time:</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="gp">>>> </span><span class="n">indicator</span><span class="o">.</span><span class="n">features_</span>
<span class="go">array([0, 1, 3])</span>
</pre></div>
</div>
<p>The <code class="docutils literal notranslate"><span class="pre">features</span></code> parameter can be set to <code class="docutils literal notranslate"><span class="pre">'all'</span></code> to return all features
whether or not they contain missing values:</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="gp">>>> </span><span class="n">indicator</span> <span class="o">=</span> <span class="n">MissingIndicator</span><span class="p">(</span><span class="n">missing_values</span><span class="o">=-</span><span class="mi">1</span><span class="p">,</span> <span class="n">features</span><span class="o">=</span><span class="s2">"all"</span><span class="p">)</span>
<span class="gp">>>> </span><span class="n">mask_all</span> <span class="o">=</span> <span class="n">indicator</span><span class="o">.</span><span class="n">fit_transform</span><span class="p">(</span><span class="n">X</span><span class="p">)</span>
<span class="gp">>>> </span><span class="n">mask_all</span>
<span class="go">array([[ True, True, False, False],</span>
<span class="go"> [False, True, False, True],</span>
<span class="go"> [False, True, False, False]])</span>
<span class="gp">>>> </span><span class="n">indicator</span><span class="o">.</span><span class="n">features_</span>
<span class="go">array([0, 1, 2, 3])</span>
</pre></div>
</div>
<p>When using the <a class="reference internal" href="generated/sklearn.impute.MissingIndicator.html#sklearn.impute.MissingIndicator" title="sklearn.impute.MissingIndicator"><code class="xref py py-class docutils literal notranslate"><span class="pre">MissingIndicator</span></code></a> in a
<a class="reference internal" href="generated/sklearn.pipeline.Pipeline.html#sklearn.pipeline.Pipeline" title="sklearn.pipeline.Pipeline"><code class="xref py py-class docutils literal notranslate"><span class="pre">Pipeline</span></code></a>, be sure to use the
<a class="reference internal" href="generated/sklearn.pipeline.FeatureUnion.html#sklearn.pipeline.FeatureUnion" title="sklearn.pipeline.FeatureUnion"><code class="xref py py-class docutils literal notranslate"><span class="pre">FeatureUnion</span></code></a> or
<a class="reference internal" href="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> to add the indicator features to
the regular features. First we obtain the <code class="docutils literal notranslate"><span class="pre">iris</span></code> dataset, and add some missing
values to it.</p>
<div class="doctest highlight-default notranslate"><div class="highlight"><pre><span></span><span class="gp">>>> </span><span class="kn">from</span> <span class="nn">sklearn.datasets</span> <span class="kn">import</span> <span class="n">load_iris</span>
<span class="gp">>>> </span><span class="kn">from</span> <span class="nn">sklearn.impute</span> <span class="kn">import</span> <span class="n">SimpleImputer</span><span class="p">,</span> <span class="n">MissingIndicator</span>
<span class="gp">>>> </span><span class="kn">from</span> <span class="nn">sklearn.model_selection</span> <span class="kn">import</span> <span class="n">train_test_split</span>
<span class="gp">>>> </span><span class="kn">from</span> <span class="nn">sklearn.pipeline</span> <span class="kn">import</span> <span class="n">FeatureUnion</span><span class="p">,</span> <span class="n">make_pipeline</span>
<span class="gp">>>> </span><span class="kn">from</span> <span class="nn">sklearn.tree</span> <span class="kn">import</span> <span class="n">DecisionTreeClassifier</span>
<span class="gp">>>> </span><span class="n">X</span><span class="p">,</span> <span class="n">y</span> <span class="o">=</span> <span class="n">load_iris</span><span class="p">(</span><span class="n">return_X_y</span><span class="o">=</span><span class="kc">True</span><span class="p">)</span>
<span class="gp">>>> </span><span class="n">mask</span> <span class="o">=</span> <span class="n">np</span><span class="o">.</span><span class="n">random</span><span class="o">.</span><span class="n">randint</span><span class="p">(</span><span class="mi">0</span><span class="p">,</span> <span class="mi">2</span><span class="p">,</span> <span class="n">size</span><span class="o">=</span><span class="n">X</span><span class="o">.</span><span class="n">shape</span><span class="p">)</span><span class="o">.</span><span class="n">astype</span><span class="p">(</span><span class="nb">bool</span><span class="p">)</span>
<span class="gp">>>> </span><span class="n">X</span><span class="p">[</span><span class="n">mask</span><span class="p">]</span> <span class="o">=</span> <span class="n">np</span><span class="o">.</span><span class="n">nan</span>
<span class="gp">>>> </span><span class="n">X_train</span><span class="p">,</span> <span class="n">X_test</span><span class="p">,</span> <span class="n">y_train</span><span class="p">,</span> <span class="n">_</span> <span class="o">=</span> <span class="n">train_test_split</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="n">test_size</span><span class="o">=</span><span class="mi">100</span><span class="p">,</span>
<span class="gp">... </span> <span class="n">random_state</span><span class="o">=</span><span class="mi">0</span><span class="p">)</span>
</pre></div>
</div>
<p>Now we create a <a class="reference internal" href="generated/sklearn.pipeline.FeatureUnion.html#sklearn.pipeline.FeatureUnion" title="sklearn.pipeline.FeatureUnion"><code class="xref py py-class docutils literal notranslate"><span class="pre">FeatureUnion</span></code></a>. All features will be
imputed using <a class="reference internal" href="generated/sklearn.impute.SimpleImputer.html#sklearn.impute.SimpleImputer" title="sklearn.impute.SimpleImputer"><code class="xref py py-class docutils literal notranslate"><span class="pre">SimpleImputer</span></code></a>, in order to enable classifiers to work
with this data. Additionally, it adds the indicator variables from
<a class="reference internal" href="generated/sklearn.impute.MissingIndicator.html#sklearn.impute.MissingIndicator" title="sklearn.impute.MissingIndicator"><code class="xref py py-class docutils literal notranslate"><span class="pre">MissingIndicator</span></code></a>.</p>
<div class="doctest highlight-default notranslate"><div class="highlight"><pre><span></span><span class="gp">>>> </span><span class="n">transformer</span> <span class="o">=</span> <span class="n">FeatureUnion</span><span class="p">(</span>
<span class="gp">... </span> <span class="n">transformer_list</span><span class="o">=</span><span class="p">[</span>
<span class="gp">... </span> <span class="p">(</span><span class="s1">'features'</span><span class="p">,</span> <span class="n">SimpleImputer</span><span class="p">(</span><span class="n">strategy</span><span class="o">=</span><span class="s1">'mean'</span><span class="p">)),</span>
<span class="gp">... </span> <span class="p">(</span><span class="s1">'indicators'</span><span class="p">,</span> <span class="n">MissingIndicator</span><span class="p">())])</span>
<span class="gp">>>> </span><span class="n">transformer</span> <span class="o">=</span> <span class="n">transformer</span><span class="o">.</span><span class="n">fit</span><span class="p">(</span><span class="n">X_train</span><span class="p">,</span> <span class="n">y_train</span><span class="p">)</span>
<span class="gp">>>> </span><span class="n">results</span> <span class="o">=</span> <span class="n">transformer</span><span class="o">.</span><span class="n">transform</span><span class="p">(</span><span class="n">X_test</span><span class="p">)</span>
<span class="gp">>>> </span><span class="n">results</span><span class="o">.</span><span class="n">shape</span>
<span class="go">(100, 8)</span>
</pre></div>
</div>
<p>Of course, we cannot use the transformer to make any predictions. We should
wrap this in a <a class="reference internal" href="generated/sklearn.pipeline.Pipeline.html#sklearn.pipeline.Pipeline" title="sklearn.pipeline.Pipeline"><code class="xref py py-class docutils literal notranslate"><span class="pre">Pipeline</span></code></a> with a classifier (e.g., a
<a class="reference internal" href="generated/sklearn.tree.DecisionTreeClassifier.html#sklearn.tree.DecisionTreeClassifier" title="sklearn.tree.DecisionTreeClassifier"><code class="xref py py-class docutils literal notranslate"><span class="pre">DecisionTreeClassifier</span></code></a>) to be able to make predictions.</p>
<div class="doctest highlight-default notranslate"><div class="highlight"><pre><span></span><span class="gp">>>> </span><span class="n">clf</span> <span class="o">=</span> <span class="n">make_pipeline</span><span class="p">(</span><span class="n">transformer</span><span class="p">,</span> <span class="n">DecisionTreeClassifier</span><span class="p">())</span>
<span class="gp">>>> </span><span class="n">clf</span> <span class="o">=</span> <span class="n">clf</span><span class="o">.</span><span class="n">fit</span><span class="p">(</span><span class="n">X_train</span><span class="p">,</span> <span class="n">y_train</span><span class="p">)</span>
<span class="gp">>>> </span><span class="n">results</span> <span class="o">=</span> <span class="n">clf</span><span class="o">.</span><span class="n">predict</span><span class="p">(</span><span class="n">X_test</span><span class="p">)</span>
<span class="gp">>>> </span><span class="n">results</span><span class="o">.</span><span class="n">shape</span>
<span class="go">(100,)</span>
</pre></div>
</div>
</section>
<section id="estimators-that-handle-nan-values">
<h2><span class="section-number">6.4.7. </span>Estimators that handle NaN values<a class="headerlink" href="#estimators-that-handle-nan-values" title="Permalink to this heading">¶</a></h2>
<p>Some estimators are designed to handle NaN values without preprocessing.
Below is the list of these estimators, classified by type
(cluster, regressor, classifier, transform):</p>
<ul>
<li><strong>Estimators that allow NaN values for type </strong><code class="docutils literal notranslate"><span class="pre">cluster</span></code><strong>:
</strong><ul class="simple">
<li><p><a class="reference external" href="./generated/sklearn.cluster.HDBSCAN.html">HDBSCAN</a></p></li>
</ul>
</li>
<li><strong>Estimators that allow NaN values for type </strong><code class="docutils literal notranslate"><span class="pre">regressor</span></code><strong>:
</strong><ul class="simple">
<li><p><a class="reference external" href="./generated/sklearn.ensemble.BaggingRegressor.html">BaggingRegressor</a></p></li>
<li><p><a class="reference external" href="./generated/sklearn.tree.DecisionTreeRegressor.html">DecisionTreeRegressor</a></p></li>
<li><p><a class="reference external" href="./generated/sklearn.ensemble.HistGradientBoostingRegressor.html">HistGradientBoostingRegressor</a></p></li>
</ul>
</li>
<li><strong>Estimators that allow NaN values for type </strong><code class="docutils literal notranslate"><span class="pre">classifier</span></code><strong>:
</strong><ul class="simple">
<li><p><a class="reference external" href="./generated/sklearn.ensemble.BaggingClassifier.html">BaggingClassifier</a></p></li>
<li><p><a class="reference external" href="./generated/sklearn.tree.DecisionTreeClassifier.html">DecisionTreeClassifier</a></p></li>
<li><p><a class="reference external" href="./generated/sklearn.ensemble.HistGradientBoostingClassifier.html">HistGradientBoostingClassifier</a></p></li>
</ul>
</li>
<li><strong>Estimators that allow NaN values for type </strong><code class="docutils literal notranslate"><span class="pre">transformer</span></code><strong>:
</strong><ul class="simple">
<li><p><a class="reference external" href="./generated/sklearn.impute.IterativeImputer.html">IterativeImputer</a></p></li>
<li><p><a class="reference external" href="./generated/sklearn.impute.KNNImputer.html">KNNImputer</a></p></li>
<li><p><a class="reference external" href="./generated/sklearn.preprocessing.MaxAbsScaler.html">MaxAbsScaler</a></p></li>
<li><p><a class="reference external" href="./generated/sklearn.preprocessing.MinMaxScaler.html">MinMaxScaler</a></p></li>
<li><p><a class="reference external" href="./generated/sklearn.impute.MissingIndicator.html">MissingIndicator</a></p></li>
<li><p><a class="reference external" href="./generated/sklearn.preprocessing.PowerTransformer.html">PowerTransformer</a></p></li>
<li><p><a class="reference external" href="./generated/sklearn.preprocessing.QuantileTransformer.html">QuantileTransformer</a></p></li>
<li><p><a class="reference external" href="./generated/sklearn.preprocessing.RobustScaler.html">RobustScaler</a></p></li>
<li><p><a class="reference external" href="./generated/sklearn.impute.SimpleImputer.html">SimpleImputer</a></p></li>
<li><p><a class="reference external" href="./generated/sklearn.preprocessing.StandardScaler.html">StandardScaler</a></p></li>
<li><p><a class="reference external" href="./generated/sklearn.feature_selection.VarianceThreshold.html">VarianceThreshold</a></p></li>
</ul>
</li>
</ul>
</section>
</section>
</div>
<div class="container">
<footer class="sk-content-footer">
© 2007 - 2023, scikit-learn developers (BSD License).
<a href="../_sources/modules/impute.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>