forked from wizardforcel/pandas-doc-zh
-
Notifications
You must be signed in to change notification settings - Fork 2
Expand file tree
/
Copy pathmissing_data.html
More file actions
1154 lines (1070 loc) · 100 KB
/
missing_data.html
File metadata and controls
1154 lines (1070 loc) · 100 KB
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
<span id="missing-data"></span><h1><span class="yiyi-st" id="yiyi-64">Working with missing data</span></h1>
<blockquote>
<p>原文：<a href="http://pandas.pydata.org/pandas-docs/stable/missing_data.html">http://pandas.pydata.org/pandas-docs/stable/missing_data.html</a></p>
<p>译者：<a href="https://github.com/wizardforcel">飞龙</a> <a href="http://usyiyi.cn/">UsyiyiCN</a></p>
<p>校对：（虚位以待）</p>
</blockquote>
<p><span class="yiyi-st" id="yiyi-65">在本节中，我们将讨论pandas中的缺失（也称为NA）值。</span></p>
<div class="admonition note">
<p class="first admonition-title"><span class="yiyi-st" id="yiyi-66">注意</span></p>
<p class="last"><span class="yiyi-st" id="yiyi-67">在内部使用<code class="docutils literal"><span class="pre">NaN</span></code>表示丢失数据的选择在很大程度上是出于简单性和性能原因。</span><span class="yiyi-st" id="yiyi-68">它与MaskedArray方法不同，例如<code class="xref py py-mod docutils literal"><span class="pre">scikits.timeseries</span></code>。</span><span class="yiyi-st" id="yiyi-69">我们希望NumPy很快能够提供一个原生NA类型的解决方案（类似于R）的性能足以用于熊猫。</span></p>
</div>
<p><span class="yiyi-st" id="yiyi-70">有关某些高级策略，请参阅<a class="reference internal" href="cookbook.html#cookbook-missing-data"><span class="std std-ref">cookbook</span></a></span></p>
<div class="section" id="missing-data-basics">
<h2><span class="yiyi-st" id="yiyi-71">Missing data basics</span></h2>
<div class="section" id="when-why-does-data-become-missing">
<h3><span class="yiyi-st" id="yiyi-72">When / why does data become missing?</span></h3>
<p><span class="yiyi-st" id="yiyi-73">有些人可能会对我们使用<em>缺少的</em>产生疑惑。</span><span class="yiyi-st" id="yiyi-74">“缺少”我们只是指<strong>null</strong>或“不存在为什么原因”。</span><span class="yiyi-st" id="yiyi-75">许多数据集只是带有缺失数据到达，或者是因为它存在并且没有被收集或者它从来不存在。</span><span class="yiyi-st" id="yiyi-76">例如，在财务时间系列的集合中，某些时间系列可能在不同的日期开始。</span><span class="yiyi-st" id="yiyi-77">因此，在开始日期之前的值通常将被标记为缺失。</span></p>
<p><span class="yiyi-st" id="yiyi-78">在pandas中，丢失数据<strong>引入</strong>到数据集中的最常见方法之一是通过重新索引。</span><span class="yiyi-st" id="yiyi-79">例如</span></p>
<div class="highlight-ipython"><div class="highlight"><pre><span></span><span class="gp">In [1]: </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="n">np</span><span class="o">.</span><span class="n">random</span><span class="o">.</span><span class="n">randn</span><span class="p">(</span><span class="mi">5</span><span class="p">,</span> <span class="mi">3</span><span class="p">),</span> <span class="n">index</span><span class="o">=</span><span class="p">[</span><span class="s1">'a'</span><span class="p">,</span> <span class="s1">'c'</span><span class="p">,</span> <span class="s1">'e'</span><span class="p">,</span> <span class="s1">'f'</span><span class="p">,</span> <span class="s1">'h'</span><span class="p">],</span>
<span class="gp"> ...:</span> <span class="n">columns</span><span class="o">=</span><span class="p">[</span><span class="s1">'one'</span><span class="p">,</span> <span class="s1">'two'</span><span class="p">,</span> <span class="s1">'three'</span><span class="p">])</span>
<span class="gp"> ...:</span>
<span class="gp">In [2]: </span><span class="n">df</span><span class="p">[</span><span class="s1">'four'</span><span class="p">]</span> <span class="o">=</span> <span class="s1">'bar'</span>
<span class="gp">In [3]: </span><span class="n">df</span><span class="p">[</span><span class="s1">'five'</span><span class="p">]</span> <span class="o">=</span> <span class="n">df</span><span class="p">[</span><span class="s1">'one'</span><span class="p">]</span> <span class="o">></span> <span class="mi">0</span>
<span class="gp">In [4]: </span><span class="n">df</span>
<span class="gr">Out[4]: </span>
<span class="go"> one two three four five</span>
<span class="go">a 0.469112 -0.282863 -1.509059 bar True</span>
<span class="go">c -1.135632 1.212112 -0.173215 bar False</span>
<span class="go">e 0.119209 -1.044236 -0.861849 bar True</span>
<span class="go">f -2.104569 -0.494929 1.071804 bar False</span>
<span class="go">h 0.721555 -0.706771 -1.039575 bar True</span>
<span class="gp">In [5]: </span><span class="n">df2</span> <span class="o">=</span> <span class="n">df</span><span class="o">.</span><span class="n">reindex</span><span class="p">([</span><span class="s1">'a'</span><span class="p">,</span> <span class="s1">'b'</span><span class="p">,</span> <span class="s1">'c'</span><span class="p">,</span> <span class="s1">'d'</span><span class="p">,</span> <span class="s1">'e'</span><span class="p">,</span> <span class="s1">'f'</span><span class="p">,</span> <span class="s1">'g'</span><span class="p">,</span> <span class="s1">'h'</span><span class="p">])</span>
<span class="gp">In [6]: </span><span class="n">df2</span>
<span class="gr">Out[6]: </span>
<span class="go"> one two three four five</span>
<span class="go">a 0.469112 -0.282863 -1.509059 bar True</span>
<span class="go">b NaN NaN NaN NaN NaN</span>
<span class="go">c -1.135632 1.212112 -0.173215 bar False</span>
<span class="go">d NaN NaN NaN NaN NaN</span>
<span class="go">e 0.119209 -1.044236 -0.861849 bar True</span>
<span class="go">f -2.104569 -0.494929 1.071804 bar False</span>
<span class="go">g NaN NaN NaN NaN NaN</span>
<span class="go">h 0.721555 -0.706771 -1.039575 bar True</span>
</pre></div>
</div>
</div>
<div class="section" id="values-considered-missing">
<h3><span class="yiyi-st" id="yiyi-80">Values considered “missing”</span></h3>
<p><span class="yiyi-st" id="yiyi-81">由于数据有多种形式和形式，pandas旨在灵活处理丢失的数据。</span><span class="yiyi-st" id="yiyi-82">由于计算速度和方便性的原因，<code class="docutils literal"><span class="pre">NaN</span></code>是默认的缺失值标记，我们需要能够使用不同类型的数据轻松检测此值：浮点，整数，布尔值和常规对象。</span><span class="yiyi-st" id="yiyi-83">然而，在许多情况下，Python <code class="docutils literal"><span class="pre">None</span></code>将出现，我们希望也考虑“missing”或“null”。</span></p>
<div class="admonition note">
<p class="first admonition-title"><span class="yiyi-st" id="yiyi-84">注意</span></p>
<p class="last"><span class="yiyi-st" id="yiyi-85">在版本v0.10.0 <code class="docutils literal"><span class="pre">inf</span></code>和<code class="docutils literal"><span class="pre">-inf</span></code>之前，在计算中也被认为是“null”。</span><span class="yiyi-st" id="yiyi-86">默认情况下不再是这种情况；请使用<code class="docutils literal"><span class="pre">mode.use_inf_as_null</span></code>选项恢复它。</span></p>
</div>
<p id="missing-isnull"><span class="yiyi-st" id="yiyi-87">pandas为了更好的处理缺失值(包括不同列的不同类型), 提供了 <code class="xref py py-func docutils literal"><span class="pre">isnull()</span></code> 和 <code class="xref py py-func docutils literal"><span class="pre">notnull()</span></code> 函数, 这两种方法都可以用在 <code class="docutils literal"><span class="pre">Series</span></code> 和 <code class="docutils literal"><span class="pre">DataFrame</span></code> 对象上:</span></p>
<div class="highlight-ipython"><div class="highlight"><pre><span></span><span class="gp">In [7]: </span><span class="n">df2</span><span class="p">[</span><span class="s1">'one'</span><span class="p">]</span>
<span class="gr">Out[7]: </span>
<span class="go">a 0.469112</span>
<span class="go">b NaN</span>
<span class="go">c -1.135632</span>
<span class="go">d NaN</span>
<span class="go">e 0.119209</span>
<span class="go">f -2.104569</span>
<span class="go">g NaN</span>
<span class="go">h 0.721555</span>
<span class="go">Name: one, dtype: float64</span>
<span class="gp">In [8]: </span><span class="n">pd</span><span class="o">.</span><span class="n">isnull</span><span class="p">(</span><span class="n">df2</span><span class="p">[</span><span class="s1">'one'</span><span class="p">])</span>
<span class="gr">Out[8]: </span>
<span class="go">a False</span>
<span class="go">b True</span>
<span class="go">c False</span>
<span class="go">d True</span>
<span class="go">e False</span>
<span class="go">f False</span>
<span class="go">g True</span>
<span class="go">h False</span>
<span class="go">Name: one, dtype: bool</span>
<span class="gp">In [9]: </span><span class="n">df2</span><span class="p">[</span><span class="s1">'four'</span><span class="p">]</span><span class="o">.</span><span class="n">notnull</span><span class="p">()</span>
<span class="gr">Out[9]: </span>
<span class="go">a True</span>
<span class="go">b False</span>
<span class="go">c True</span>
<span class="go">d False</span>
<span class="go">e True</span>
<span class="go">f True</span>
<span class="go">g False</span>
<span class="go">h True</span>
<span class="go">Name: four, dtype: bool</span>
<span class="gp">In [10]: </span><span class="n">df2</span><span class="o">.</span><span class="n">isnull</span><span class="p">()</span>
<span class="gr">Out[10]: </span>
<span class="go"> one two three four five</span>
<span class="go">a False False False False False</span>
<span class="go">b True True True True True</span>
<span class="go">c False False False False False</span>
<span class="go">d True True True True True</span>
<span class="go">e False False False False False</span>
<span class="go">f False False False False False</span>
<span class="go">g True True True True True</span>
<span class="go">h False False False False False</span>
</pre></div>
</div>
<div class="admonition warning">
<p class="first admonition-title"><span class="yiyi-st" id="yiyi-88">警告</span></p>
<p><span class="yiyi-st" id="yiyi-89">必须注意，在python（和numpy）中，<code class="docutils literal"><span class="pre">nan's</span></code>不比较相等，但<code class="docutils literal"><span class="pre">None's</span></code> <strong>do</strong>。</span><span class="yiyi-st" id="yiyi-90">Note that Pandas/numpy uses the fact that <code class="docutils literal"><span class="pre">np.nan</span> <span class="pre">!=</span> <span class="pre">np.nan</span></code>, and treats <code class="docutils literal"><span class="pre">None</span></code> like <code class="docutils literal"><span class="pre">np.nan</span></code>.</span></p>
<div class="highlight-ipython"><div class="highlight"><pre><span></span><span class="gp">In [11]: </span><span class="bp">None</span> <span class="o">==</span> <span class="bp">None</span>
<span class="gr">Out[11]: </span><span class="bp">True</span>
<span class="gp">In [12]: </span><span class="n">np</span><span class="o">.</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="gr">Out[12]: </span><span class="bp">False</span>
</pre></div>
</div>
<p><span class="yiyi-st" id="yiyi-91">因此，与上述相比，标量等式比较与<code class="docutils literal"><span class="pre">None/np.nan</span></code>不提供有用的信息。</span></p>
<div class="last highlight-ipython"><div class="highlight"><pre><span></span><span class="gp">In [13]: </span><span class="n">df2</span><span class="p">[</span><span class="s1">'one'</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="gr">Out[13]: </span>
<span class="go">a False</span>
<span class="go">b False</span>
<span class="go">c False</span>
<span class="go">d False</span>
<span class="go">e False</span>
<span class="go">f False</span>
<span class="go">g False</span>
<span class="go">h False</span>
<span class="go">Name: one, dtype: bool</span>
</pre></div>
</div>
</div>
</div>
</div>
<div class="section" id="datetimes">
<h2><span class="yiyi-st" id="yiyi-92">Datetimes</span></h2>
<p><span class="yiyi-st" id="yiyi-93">对于datetime64 [ns]类型，<code class="docutils literal"><span class="pre">NaT</span></code>表示缺少的值。</span><span class="yiyi-st" id="yiyi-94">这是一个伪本地的哨兵值，可以用单数dtype（datetime64 [ns]）中的numpy表示。</span><span class="yiyi-st" id="yiyi-95">pandas对象提供<code class="docutils literal"><span class="pre">NaT</span></code>和<code class="docutils literal"><span class="pre">NaN</span></code>之间的相互兼容性。</span></p>
<div class="highlight-ipython"><div class="highlight"><pre><span></span><span class="gp">In [14]: </span><span class="n">df2</span> <span class="o">=</span> <span class="n">df</span><span class="o">.</span><span class="n">copy</span><span class="p">()</span>
<span class="gp">In [15]: </span><span class="n">df2</span><span class="p">[</span><span class="s1">'timestamp'</span><span class="p">]</span> <span class="o">=</span> <span class="n">pd</span><span class="o">.</span><span class="n">Timestamp</span><span class="p">(</span><span class="s1">'20120101'</span><span class="p">)</span>
<span class="gp">In [16]: </span><span class="n">df2</span>
<span class="gr">Out[16]: </span>
<span class="go"> one two three four five timestamp</span>
<span class="go">a 0.469112 -0.282863 -1.509059 bar True 2012-01-01</span>
<span class="go">c -1.135632 1.212112 -0.173215 bar False 2012-01-01</span>
<span class="go">e 0.119209 -1.044236 -0.861849 bar True 2012-01-01</span>
<span class="go">f -2.104569 -0.494929 1.071804 bar False 2012-01-01</span>
<span class="go">h 0.721555 -0.706771 -1.039575 bar True 2012-01-01</span>
<span class="gp">In [17]: </span><span class="n">df2</span><span class="o">.</span><span class="n">ix</span><span class="p">[[</span><span class="s1">'a'</span><span class="p">,</span><span class="s1">'c'</span><span class="p">,</span><span class="s1">'h'</span><span class="p">],[</span><span class="s1">'one'</span><span class="p">,</span><span class="s1">'timestamp'</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">In [18]: </span><span class="n">df2</span>
<span class="gr">Out[18]: </span>
<span class="go"> one two three four five timestamp</span>
<span class="go">a NaN -0.282863 -1.509059 bar True NaT</span>
<span class="go">c NaN 1.212112 -0.173215 bar False NaT</span>
<span class="go">e 0.119209 -1.044236 -0.861849 bar True 2012-01-01</span>
<span class="go">f -2.104569 -0.494929 1.071804 bar False 2012-01-01</span>
<span class="go">h NaN -0.706771 -1.039575 bar True NaT</span>
<span class="gp">In [19]: </span><span class="n">df2</span><span class="o">.</span><span class="n">get_dtype_counts</span><span class="p">()</span>
<span class="gr">Out[19]: </span>
<span class="go">bool 1</span>
<span class="go">datetime64[ns] 1</span>
<span class="go">float64 3</span>
<span class="go">object 1</span>
<span class="go">dtype: int64</span>
</pre></div>
</div>
</div>
<div class="section" id="inserting-missing-data">
<span id="missing-inserting"></span><h2><span class="yiyi-st" id="yiyi-96">Inserting missing data</span></h2>
<p><span class="yiyi-st" id="yiyi-97">您可以通过简单地分配到容器来插入缺失值。</span><span class="yiyi-st" id="yiyi-98">使用的实际缺失值将基于dtype进行选择。</span></p>
<p><span class="yiyi-st" id="yiyi-99">例如，无论选择的缺少值类型如何，数值型将始终使用<code class="docutils literal"><span class="pre">NaN</span></code>：</span></p>
<div class="highlight-ipython"><div class="highlight"><pre><span></span><span class="gp">In [20]: </span><span class="n">s</span> <span class="o">=</span> <span class="n">pd</span><span class="o">.</span><span class="n">Series</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="mi">3</span><span class="p">])</span>
<span class="gp">In [21]: </span><span class="n">s</span><span class="o">.</span><span class="n">loc</span><span class="p">[</span><span class="mi">0</span><span class="p">]</span> <span class="o">=</span> <span class="bp">None</span>
<span class="gp">In [22]: </span><span class="n">s</span>
<span class="gr">Out[22]: </span>
<span class="go">0 NaN</span>
<span class="go">1 2.0</span>
<span class="go">2 3.0</span>
<span class="go">dtype: float64</span>
</pre></div>
</div>
<p><span class="yiyi-st" id="yiyi-100">同样，datetime对象将始终使用<code class="docutils literal"><span class="pre">NaT</span></code>。</span></p>
<p><span class="yiyi-st" id="yiyi-101">对于object类型，pandas将使用给定的值：</span></p>
<div class="highlight-ipython"><div class="highlight"><pre><span></span><span class="gp">In [23]: </span><span class="n">s</span> <span class="o">=</span> <span class="n">pd</span><span class="o">.</span><span class="n">Series</span><span class="p">([</span><span class="s2">"a"</span><span class="p">,</span> <span class="s2">"b"</span><span class="p">,</span> <span class="s2">"c"</span><span class="p">])</span>
<span class="gp">In [24]: </span><span class="n">s</span><span class="o">.</span><span class="n">loc</span><span class="p">[</span><span class="mi">0</span><span class="p">]</span> <span class="o">=</span> <span class="bp">None</span>
<span class="gp">In [25]: </span><span class="n">s</span><span class="o">.</span><span class="n">loc</span><span class="p">[</span><span class="mi">1</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">In [26]: </span><span class="n">s</span>
<span class="gr">Out[26]: </span>
<span class="go">0 None</span>
<span class="go">1 NaN</span>
<span class="go">2 c</span>
<span class="go">dtype: object</span>
</pre></div>
</div>
</div>
<div class="section" id="calculations-with-missing-data">
<h2><span class="yiyi-st" id="yiyi-102">Calculations with missing data</span></h2>
<p><span class="yiyi-st" id="yiyi-103">缺失值通过pandas对象之间的算术运算自然传播。</span></p>
<div class="highlight-ipython"><div class="highlight"><pre><span></span><span class="gp">In [27]: </span><span class="n">a</span>
<span class="gr">Out[27]: </span>
<span class="go"> one two</span>
<span class="go">a NaN -0.282863</span>
<span class="go">c NaN 1.212112</span>
<span class="go">e 0.119209 -1.044236</span>
<span class="go">f -2.104569 -0.494929</span>
<span class="go">h -2.104569 -0.706771</span>
<span class="gp">In [28]: </span><span class="n">b</span>
<span class="gr">Out[28]: </span>
<span class="go"> one two three</span>
<span class="go">a NaN -0.282863 -1.509059</span>
<span class="go">c NaN 1.212112 -0.173215</span>
<span class="go">e 0.119209 -1.044236 -0.861849</span>
<span class="go">f -2.104569 -0.494929 1.071804</span>
<span class="go">h NaN -0.706771 -1.039575</span>
<span class="gp">In [29]: </span><span class="n">a</span> <span class="o">+</span> <span class="n">b</span>
<span class="gr">Out[29]: </span>
<span class="go"> one three two</span>
<span class="go">a NaN NaN -0.565727</span>
<span class="go">c NaN NaN 2.424224</span>
<span class="go">e 0.238417 NaN -2.088472</span>
<span class="go">f -4.209138 NaN -0.989859</span>
<span class="go">h NaN NaN -1.413542</span>
</pre></div>
</div>
<p><span class="yiyi-st" id="yiyi-104">在<a class="reference internal" href="basics.html#basics-stats"><span class="std std-ref">data structure overview</span></a>（此处列出<a class="reference internal" href="api.html#api-series-stats"><span class="std std-ref">here</span></a>和<a class="reference internal" href="api.html#api-dataframe-stats"><span class="std std-ref">here</span></a>）中讨论的描述性统计和计算方法都是为了记录丢失的数据。</span><span class="yiyi-st" id="yiyi-105">例如：</span></p>
<ul class="simple">
<li><span class="yiyi-st" id="yiyi-106">当对数据求和时，NA（缺失）值将被视为零</span></li>
<li><span class="yiyi-st" id="yiyi-107">如果数据都是NA，则结果将是NA</span></li>
<li><span class="yiyi-st" id="yiyi-108"><strong>cumsum</strong>和<strong>cumprod</strong>等方法忽略NA值，但在生成的数组中保留它们</span></li>
</ul>
<div class="highlight-ipython"><div class="highlight"><pre><span></span><span class="gp">In [30]: </span><span class="n">df</span>
<span class="gr">Out[30]: </span>
<span class="go"> one two three</span>
<span class="go">a NaN -0.282863 -1.509059</span>
<span class="go">c NaN 1.212112 -0.173215</span>
<span class="go">e 0.119209 -1.044236 -0.861849</span>
<span class="go">f -2.104569 -0.494929 1.071804</span>
<span class="go">h NaN -0.706771 -1.039575</span>
<span class="gp">In [31]: </span><span class="n">df</span><span class="p">[</span><span class="s1">'one'</span><span class="p">]</span><span class="o">.</span><span class="n">sum</span><span class="p">()</span>
<span class="gr">Out[31]: </span><span class="o">-</span><span class="mf">1.9853605075978744</span>
<span class="gp">In [32]: </span><span class="n">df</span><span class="o">.</span><span class="n">mean</span><span class="p">(</span><span class="mi">1</span><span class="p">)</span>
<span class="gr">Out[32]: </span>
<span class="go">a -0.895961</span>
<span class="go">c 0.519449</span>
<span class="go">e -0.595625</span>
<span class="go">f -0.509232</span>
<span class="go">h -0.873173</span>
<span class="go">dtype: float64</span>
<span class="gp">In [33]: </span><span class="n">df</span><span class="o">.</span><span class="n">cumsum</span><span class="p">()</span>
<span class="gr">Out[33]: </span>
<span class="go"> one two three</span>
<span class="go">a NaN -0.282863 -1.509059</span>
<span class="go">c NaN 0.929249 -1.682273</span>
<span class="go">e 0.119209 -0.114987 -2.544122</span>
<span class="go">f -1.985361 -0.609917 -1.472318</span>
<span class="go">h NaN -1.316688 -2.511893</span>
</pre></div>
</div>
<div class="section" id="na-values-in-groupby">
<h3><span class="yiyi-st" id="yiyi-109">NA values in GroupBy</span></h3>
<p><span class="yiyi-st" id="yiyi-110">GroupBy中的NA组将自动排除。</span><span class="yiyi-st" id="yiyi-111">此行为与R一致，例如：</span></p>
<div class="highlight-ipython"><div class="highlight"><pre><span></span><span class="gp">In [34]: </span><span class="n">df</span>
<span class="gr">Out[34]: </span>
<span class="go"> one two three</span>
<span class="go">a NaN -0.282863 -1.509059</span>
<span class="go">c NaN 1.212112 -0.173215</span>
<span class="go">e 0.119209 -1.044236 -0.861849</span>
<span class="go">f -2.104569 -0.494929 1.071804</span>
<span class="go">h NaN -0.706771 -1.039575</span>
<span class="gp">In [35]: </span><span class="n">df</span><span class="o">.</span><span class="n">groupby</span><span class="p">(</span><span class="s1">'one'</span><span class="p">)</span><span class="o">.</span><span class="n">mean</span><span class="p">()</span>
<span class="gr">Out[35]: </span>
<span class="go"> two three</span>
<span class="go">one </span>
<span class="go">-2.104569 -0.494929 1.071804</span>
<span class="go"> 0.119209 -1.044236 -0.861849</span>
</pre></div>
</div>
<p><span class="yiyi-st" id="yiyi-112">有关详细信息，请参阅<a class="reference internal" href="groupby.html#groupby-missing"><span class="std std-ref">here</span></a>部分。</span></p>
</div>
</div>
<div class="section" id="cleaning-filling-missing-data">
<h2><span class="yiyi-st" id="yiyi-113">Cleaning / filling missing data</span></h2>
<p><span class="yiyi-st" id="yiyi-114">pandas对象配备了各种数据处理方法来处理丢失的数据。</span></p>
<div class="section" id="filling-missing-values-fillna">
<span id="missing-data-fillna"></span><h3><span class="yiyi-st" id="yiyi-115">Filling missing values: fillna</span></h3>
<p><span class="yiyi-st" id="yiyi-116"><strong>fillna</strong>函数可以通过两种方式“填充”NA值与非空数据，我们说明：</span></p>
<p><span class="yiyi-st" id="yiyi-117"><strong>将NA替换为标量值</strong></span></p>
<div class="highlight-ipython"><div class="highlight"><pre><span></span><span class="gp">In [36]: </span><span class="n">df2</span>
<span class="gr">Out[36]: </span>
<span class="go"> one two three four five timestamp</span>
<span class="go">a NaN -0.282863 -1.509059 bar True NaT</span>
<span class="go">c NaN 1.212112 -0.173215 bar False NaT</span>
<span class="go">e 0.119209 -1.044236 -0.861849 bar True 2012-01-01</span>
<span class="go">f -2.104569 -0.494929 1.071804 bar False 2012-01-01</span>
<span class="go">h NaN -0.706771 -1.039575 bar True NaT</span>
<span class="gp">In [37]: </span><span class="n">df2</span><span class="o">.</span><span class="n">fillna</span><span class="p">(</span><span class="mi">0</span><span class="p">)</span>
<span class="gr">Out[37]: </span>
<span class="go"> one two three four five timestamp</span>
<span class="go">a 0.000000 -0.282863 -1.509059 bar True 1970-01-01</span>
<span class="go">c 0.000000 1.212112 -0.173215 bar False 1970-01-01</span>
<span class="go">e 0.119209 -1.044236 -0.861849 bar True 2012-01-01</span>
<span class="go">f -2.104569 -0.494929 1.071804 bar False 2012-01-01</span>
<span class="go">h 0.000000 -0.706771 -1.039575 bar True 1970-01-01</span>
<span class="gp">In [38]: </span><span class="n">df2</span><span class="p">[</span><span class="s1">'four'</span><span class="p">]</span><span class="o">.</span><span class="n">fillna</span><span class="p">(</span><span class="s1">'missing'</span><span class="p">)</span>
<span class="gr">Out[38]: </span>
<span class="go">a bar</span>
<span class="go">c bar</span>
<span class="go">e bar</span>
<span class="go">f bar</span>
<span class="go">h bar</span>
<span class="go">Name: four, dtype: object</span>
</pre></div>
</div>
<p><span class="yiyi-st" id="yiyi-118"><strong>向前或向后填充间隙</strong></span></p>
<p><span class="yiyi-st" id="yiyi-119">使用与<a class="reference internal" href="basics.html#basics-reindexing"><span class="std std-ref">reindexing</span></a>相同的填充参数，我们可以向前或向后传播非空值：</span></p>
<div class="highlight-ipython"><div class="highlight"><pre><span></span><span class="gp">In [39]: </span><span class="n">df</span>
<span class="gr">Out[39]: </span>
<span class="go"> one two three</span>
<span class="go">a NaN -0.282863 -1.509059</span>
<span class="go">c NaN 1.212112 -0.173215</span>
<span class="go">e 0.119209 -1.044236 -0.861849</span>
<span class="go">f -2.104569 -0.494929 1.071804</span>
<span class="go">h NaN -0.706771 -1.039575</span>
<span class="gp">In [40]: </span><span class="n">df</span><span class="o">.</span><span class="n">fillna</span><span class="p">(</span><span class="n">method</span><span class="o">=</span><span class="s1">'pad'</span><span class="p">)</span>
<span class="gr">Out[40]: </span>
<span class="go"> one two three</span>
<span class="go">a NaN -0.282863 -1.509059</span>
<span class="go">c NaN 1.212112 -0.173215</span>
<span class="go">e 0.119209 -1.044236 -0.861849</span>
<span class="go">f -2.104569 -0.494929 1.071804</span>
<span class="go">h -2.104569 -0.706771 -1.039575</span>
</pre></div>
</div>
<p id="missing-data-fillna-limit"><span class="yiyi-st" id="yiyi-120"><strong>控制填充的缺失值数量</strong></span></p>
<p><span class="yiyi-st" id="yiyi-121">如果我们只想让连续的间隙填充到一定数量的数据点，我们可以使用<cite>limit</cite>关键字：</span></p>
<div class="highlight-ipython"><div class="highlight"><pre><span></span><span class="gp">In [41]: </span><span class="n">df</span>
<span class="gr">Out[41]: </span>
<span class="go"> one two three</span>
<span class="go">a NaN -0.282863 -1.509059</span>
<span class="go">c NaN 1.212112 -0.173215</span>
<span class="go">e NaN NaN NaN</span>
<span class="go">f NaN NaN NaN</span>
<span class="go">h NaN -0.706771 -1.039575</span>
<span class="gp">In [42]: </span><span class="n">df</span><span class="o">.</span><span class="n">fillna</span><span class="p">(</span><span class="n">method</span><span class="o">=</span><span class="s1">'pad'</span><span class="p">,</span> <span class="n">limit</span><span class="o">=</span><span class="mi">1</span><span class="p">)</span>
<span class="gr">Out[42]: </span>
<span class="go"> one two three</span>
<span class="go">a NaN -0.282863 -1.509059</span>
<span class="go">c NaN 1.212112 -0.173215</span>
<span class="go">e NaN 1.212112 -0.173215</span>
<span class="go">f NaN NaN NaN</span>
<span class="go">h NaN -0.706771 -1.039575</span>
</pre></div>
</div>
<p><span class="yiyi-st" id="yiyi-122">为了提醒您，这些是可用的填充方法：</span></p>
<table border="1" class="docutils">
<colgroup>
<col width="38%">
<col width="63%">
</colgroup>
<thead valign="bottom">
<tr class="row-odd"><th class="head"><span class="yiyi-st" id="yiyi-123">方法</span></th>
<th class="head"><span class="yiyi-st" id="yiyi-124">行动</span></th>
</tr>
</thead>
<tbody valign="top">
<tr class="row-even"><td><span class="yiyi-st" id="yiyi-125">pad / ffill</span></td>
<td><span class="yiyi-st" id="yiyi-126">向前填充值</span></td>
</tr>
<tr class="row-odd"><td><span class="yiyi-st" id="yiyi-127">bfill / backfill</span></td>
<td><span class="yiyi-st" id="yiyi-128">向后填充值</span></td>
</tr>
</tbody>
</table>
<p><span class="yiyi-st" id="yiyi-129">使用时间序列数据，使用pad / ffill非常常见，因此“最后已知值”在每个时间点都可用。</span></p>
<p><span class="yiyi-st" id="yiyi-130"><code class="docutils literal"><span class="pre">ffill()</span></code>函数等效于<code class="docutils literal"><span class="pre">fillna(method='ffill')</span></code>和<code class="docutils literal"><span class="pre">bfill()</span></code>等效于<code class="docutils literal"><span class="pre">fillna(method='bfill')</span></code></span></p>
</div>
<div class="section" id="filling-with-a-pandasobject">
<span id="missing-data-pandasobject"></span><h3><span class="yiyi-st" id="yiyi-131">Filling with a PandasObject</span></h3>
<div class="versionadded">
<p><span class="yiyi-st" id="yiyi-132"><span class="versionmodified">版本0.12中的新功能。</span></span></p>
</div>
<p><span class="yiyi-st" id="yiyi-133">你也可以使用可以对齐的dict或者系列。</span><span class="yiyi-st" id="yiyi-134">系列的dict或index的标签必须与您要填充的框架的列匹配。</span><span class="yiyi-st" id="yiyi-135">这种情况的用法是用该列的平均值填充DataFrame。</span></p>
<div class="highlight-ipython"><div class="highlight"><pre><span></span><span class="gp">In [43]: </span><span class="n">dff</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="n">np</span><span class="o">.</span><span class="n">random</span><span class="o">.</span><span class="n">randn</span><span class="p">(</span><span class="mi">10</span><span class="p">,</span><span class="mi">3</span><span class="p">),</span> <span class="n">columns</span><span class="o">=</span><span class="nb">list</span><span class="p">(</span><span class="s1">'ABC'</span><span class="p">))</span>
<span class="gp">In [44]: </span><span class="n">dff</span><span class="o">.</span><span class="n">iloc</span><span class="p">[</span><span class="mi">3</span><span class="p">:</span><span class="mi">5</span><span class="p">,</span><span class="mi">0</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">In [45]: </span><span class="n">dff</span><span class="o">.</span><span class="n">iloc</span><span class="p">[</span><span class="mi">4</span><span class="p">:</span><span class="mi">6</span><span class="p">,</span><span class="mi">1</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">In [46]: </span><span class="n">dff</span><span class="o">.</span><span class="n">iloc</span><span class="p">[</span><span class="mi">5</span><span class="p">:</span><span class="mi">8</span><span class="p">,</span><span class="mi">2</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">In [47]: </span><span class="n">dff</span>
<span class="gr">Out[47]: </span>
<span class="go"> A B C</span>
<span class="go">0 0.271860 -0.424972 0.567020</span>
<span class="go">1 0.276232 -1.087401 -0.673690</span>
<span class="go">2 0.113648 -1.478427 0.524988</span>
<span class="go">3 NaN 0.577046 -1.715002</span>
<span class="go">4 NaN NaN -1.157892</span>
<span class="go">5 -1.344312 NaN NaN</span>
<span class="go">6 -0.109050 1.643563 NaN</span>
<span class="go">7 0.357021 -0.674600 NaN</span>
<span class="go">8 -0.968914 -1.294524 0.413738</span>
<span class="go">9 0.276662 -0.472035 -0.013960</span>
<span class="gp">In [48]: </span><span class="n">dff</span><span class="o">.</span><span class="n">fillna</span><span class="p">(</span><span class="n">dff</span><span class="o">.</span><span class="n">mean</span><span class="p">())</span>
<span class="gr">Out[48]: </span>
<span class="go"> A B C</span>
<span class="go">0 0.271860 -0.424972 0.567020</span>
<span class="go">1 0.276232 -1.087401 -0.673690</span>
<span class="go">2 0.113648 -1.478427 0.524988</span>
<span class="go">3 -0.140857 0.577046 -1.715002</span>
<span class="go">4 -0.140857 -0.401419 -1.157892</span>
<span class="go">5 -1.344312 -0.401419 -0.293543</span>
<span class="go">6 -0.109050 1.643563 -0.293543</span>
<span class="go">7 0.357021 -0.674600 -0.293543</span>
<span class="go">8 -0.968914 -1.294524 0.413738</span>
<span class="go">9 0.276662 -0.472035 -0.013960</span>
<span class="gp">In [49]: </span><span class="n">dff</span><span class="o">.</span><span class="n">fillna</span><span class="p">(</span><span class="n">dff</span><span class="o">.</span><span class="n">mean</span><span class="p">()[</span><span class="s1">'B'</span><span class="p">:</span><span class="s1">'C'</span><span class="p">])</span>
<span class="gr">Out[49]: </span>
<span class="go"> A B C</span>
<span class="go">0 0.271860 -0.424972 0.567020</span>
<span class="go">1 0.276232 -1.087401 -0.673690</span>
<span class="go">2 0.113648 -1.478427 0.524988</span>
<span class="go">3 NaN 0.577046 -1.715002</span>
<span class="go">4 NaN -0.401419 -1.157892</span>
<span class="go">5 -1.344312 -0.401419 -0.293543</span>
<span class="go">6 -0.109050 1.643563 -0.293543</span>
<span class="go">7 0.357021 -0.674600 -0.293543</span>
<span class="go">8 -0.968914 -1.294524 0.413738</span>
<span class="go">9 0.276662 -0.472035 -0.013960</span>
</pre></div>
</div>
<div class="versionadded">
<p><span class="yiyi-st" id="yiyi-136"><span class="versionmodified">版本0.13中的新功能。</span></span></p>
</div>
<p><span class="yiyi-st" id="yiyi-137">与上面的结果相同，但是对齐了“fill”值，这是一个在这种情况下的系列。</span></p>
<div class="highlight-ipython"><div class="highlight"><pre><span></span><span class="gp">In [50]: </span><span class="n">dff</span><span class="o">.</span><span class="n">where</span><span class="p">(</span><span class="n">pd</span><span class="o">.</span><span class="n">notnull</span><span class="p">(</span><span class="n">dff</span><span class="p">),</span> <span class="n">dff</span><span class="o">.</span><span class="n">mean</span><span class="p">(),</span> <span class="n">axis</span><span class="o">=</span><span class="s1">'columns'</span><span class="p">)</span>
<span class="gr">Out[50]: </span>
<span class="go"> A B C</span>
<span class="go">0 0.271860 -0.424972 0.567020</span>
<span class="go">1 0.276232 -1.087401 -0.673690</span>
<span class="go">2 0.113648 -1.478427 0.524988</span>
<span class="go">3 -0.140857 0.577046 -1.715002</span>
<span class="go">4 -0.140857 -0.401419 -1.157892</span>
<span class="go">5 -1.344312 -0.401419 -0.293543</span>
<span class="go">6 -0.109050 1.643563 -0.293543</span>
<span class="go">7 0.357021 -0.674600 -0.293543</span>
<span class="go">8 -0.968914 -1.294524 0.413738</span>
<span class="go">9 0.276662 -0.472035 -0.013960</span>
</pre></div>
</div>
</div>
<div class="section" id="dropping-axis-labels-with-missing-data-dropna">
<span id="missing-data-dropna"></span><h3><span class="yiyi-st" id="yiyi-138">Dropping axis labels with missing data: dropna</span></h3>
<p><span class="yiyi-st" id="yiyi-139">您可能希望简单地从数据集中排除涉及缺失数据的标签。</span><span class="yiyi-st" id="yiyi-140">为此，请使用<strong>dropna</strong>方法：</span></p>
<div class="highlight-ipython"><div class="highlight"><pre><span></span><span class="gp">In [51]: </span><span class="n">df</span>
<span class="gr">Out[51]: </span>
<span class="go"> one two three</span>
<span class="go">a NaN -0.282863 -1.509059</span>
<span class="go">c NaN 1.212112 -0.173215</span>
<span class="go">e NaN 0.000000 0.000000</span>
<span class="go">f NaN 0.000000 0.000000</span>
<span class="go">h NaN -0.706771 -1.039575</span>
<span class="gp">In [52]: </span><span class="n">df</span><span class="o">.</span><span class="n">dropna</span><span class="p">(</span><span class="n">axis</span><span class="o">=</span><span class="mi">0</span><span class="p">)</span>
<span class="gr">Out[52]: </span>
<span class="go">Empty DataFrame</span>
<span class="go">Columns: [one, two, three]</span>
<span class="go">Index: []</span>
<span class="gp">In [53]: </span><span class="n">df</span><span class="o">.</span><span class="n">dropna</span><span class="p">(</span><span class="n">axis</span><span class="o">=</span><span class="mi">1</span><span class="p">)</span>
<span class="gr">Out[53]: </span>
<span class="go"> two three</span>
<span class="go">a -0.282863 -1.509059</span>
<span class="go">c 1.212112 -0.173215</span>
<span class="go">e 0.000000 0.000000</span>
<span class="go">f 0.000000 0.000000</span>
<span class="go">h -0.706771 -1.039575</span>
<span class="gp">In [54]: </span><span class="n">df</span><span class="p">[</span><span class="s1">'one'</span><span class="p">]</span><span class="o">.</span><span class="n">dropna</span><span class="p">()</span>
<span class="gr">Out[54]: </span><span class="n">Series</span><span class="p">([],</span> <span class="n">Name</span><span class="p">:</span> <span class="n">one</span><span class="p">,</span> <span class="n">dtype</span><span class="p">:</span> <span class="n">float64</span><span class="p">)</span>
</pre></div>
</div>
<p><span class="yiyi-st" id="yiyi-141">Series.dropna是一个更简单的方法，因为它只有一个轴要考虑。</span><span class="yiyi-st" id="yiyi-142">DataFrame.dropna有比Series.dropna更多的选项，可以在API中检查<a class="reference internal" href="api.html#api-dataframe-missing"><span class="std std-ref">in the API</span></a></span></p>
</div>
<div class="section" id="interpolation">
<span id="missing-data-interpolate"></span><h3><span class="yiyi-st" id="yiyi-143">Interpolation</span></h3>
<div class="versionadded">
<p><span class="yiyi-st" id="yiyi-144"><span class="versionmodified">版本0.13.0中的新功能：</span> <a class="reference internal" href="generated/pandas.DataFrame.interpolate.html#pandas.DataFrame.interpolate" title="pandas.DataFrame.interpolate"><code class="xref py py-meth docutils literal"><span class="pre">interpolate()</span></code></a>和<a class="reference internal" href="generated/pandas.Series.interpolate.html#pandas.Series.interpolate" title="pandas.Series.interpolate"><code class="xref py py-meth docutils literal"><span class="pre">interpolate()</span></code></a>更新了插值方法和功能。</span></p>
</div>
<div class="versionadded">
<p><span class="yiyi-st" id="yiyi-145"><span class="versionmodified">版本0.17.0中的新功能：</span>添加了<code class="docutils literal"><span class="pre">limit_direction</span></code>关键字参数。</span></p>
</div>
<p><span class="yiyi-st" id="yiyi-146">Series和Dataframe对象都有一个<code class="docutils literal"><span class="pre">interpolate</span></code>方法，默认情况下，在缺失的数据点执行线性插值。</span></p>
<div class="highlight-ipython"><div class="highlight"><pre><span></span><span class="gp">In [55]: </span><span class="n">ts</span>
<span class="gr">Out[55]: </span>
<span class="go">2000-01-31 0.469112</span>
<span class="go">2000-02-29 NaN</span>
<span class="go">2000-03-31 NaN</span>
<span class="go">2000-04-28 NaN</span>
<span class="go">2000-05-31 NaN</span>
<span class="go">2000-06-30 NaN</span>
<span class="go">2000-07-31 NaN</span>
<span class="go"> ... </span>
<span class="go">2007-10-31 -3.305259</span>
<span class="go">2007-11-30 -5.485119</span>
<span class="go">2007-12-31 -6.854968</span>
<span class="go">2008-01-31 -7.809176</span>
<span class="go">2008-02-29 -6.346480</span>
<span class="go">2008-03-31 -8.089641</span>
<span class="go">2008-04-30 -8.916232</span>
<span class="go">Freq: BM, dtype: float64</span>
<span class="gp">In [56]: </span><span class="n">ts</span><span class="o">.</span><span class="n">count</span><span class="p">()</span>
<span class="gr">Out[56]: </span><span class="mi">61</span>
<span class="gp">In [57]: </span><span class="n">ts</span><span class="o">.</span><span class="n">interpolate</span><span class="p">()</span><span class="o">.</span><span class="n">count</span><span class="p">()</span>
<span class="gr">Out[57]: </span><span class="mi">100</span>
<span class="gp">In [58]: </span><span class="n">ts</span><span class="o">.</span><span class="n">interpolate</span><span class="p">()</span><span class="o">.</span><span class="n">plot</span><span class="p">()</span>
<span class="gr">Out[58]: </span><span class="o"><</span><span class="n">matplotlib</span><span class="o">.</span><span class="n">axes</span><span class="o">.</span><span class="n">_subplots</span><span class="o">.</span><span class="n">AxesSubplot</span> <span class="n">at</span> <span class="mh">0x7ff2667af150</span><span class="o">></span>
</pre></div>
</div>
<img alt="http://pandas.pydata.org/pandas-docs/version/0.19.2/_images/series_interpolate.png" src="http://pandas.pydata.org/pandas-docs/version/0.19.2/_images/series_interpolate.png">
<p><span class="yiyi-st" id="yiyi-147">可通过<code class="docutils literal"><span class="pre">method</span></code>关键字获得索引感知插值：</span></p>
<div class="highlight-ipython"><div class="highlight"><pre><span></span><span class="gp">In [59]: </span><span class="n">ts2</span>
<span class="gr">Out[59]: </span>
<span class="go">2000-01-31 0.469112</span>
<span class="go">2000-02-29 NaN</span>
<span class="go">2002-07-31 -5.689738</span>
<span class="go">2005-01-31 NaN</span>
<span class="go">2008-04-30 -8.916232</span>
<span class="go">dtype: float64</span>
<span class="gp">In [60]: </span><span class="n">ts2</span><span class="o">.</span><span class="n">interpolate</span><span class="p">()</span>
<span class="gr">Out[60]: </span>
<span class="go">2000-01-31 0.469112</span>
<span class="go">2000-02-29 -2.610313</span>
<span class="go">2002-07-31 -5.689738</span>
<span class="go">2005-01-31 -7.302985</span>
<span class="go">2008-04-30 -8.916232</span>
<span class="go">dtype: float64</span>
<span class="gp">In [61]: </span><span class="n">ts2</span><span class="o">.</span><span class="n">interpolate</span><span class="p">(</span><span class="n">method</span><span class="o">=</span><span class="s1">'time'</span><span class="p">)</span>
<span class="gr">Out[61]: </span>
<span class="go">2000-01-31 0.469112</span>
<span class="go">2000-02-29 0.273272</span>
<span class="go">2002-07-31 -5.689738</span>
<span class="go">2005-01-31 -7.095568</span>
<span class="go">2008-04-30 -8.916232</span>
<span class="go">dtype: float64</span>
</pre></div>
</div>
<p><span class="yiyi-st" id="yiyi-148">对于浮点索引，请使用<code class="docutils literal"><span class="pre">method='values'</span></code>：</span></p>
<div class="highlight-ipython"><div class="highlight"><pre><span></span><span class="gp">In [62]: </span><span class="n">ser</span>
<span class="gr">Out[62]: </span>
<span class="go">0.0 0.0</span>
<span class="go">1.0 NaN</span>
<span class="go">10.0 10.0</span>
<span class="go">dtype: float64</span>
<span class="gp">In [63]: </span><span class="n">ser</span><span class="o">.</span><span class="n">interpolate</span><span class="p">()</span>
<span class="gr">Out[63]: </span>
<span class="go">0.0 0.0</span>
<span class="go">1.0 5.0</span>
<span class="go">10.0 10.0</span>
<span class="go">dtype: float64</span>
<span class="gp">In [64]: </span><span class="n">ser</span><span class="o">.</span><span class="n">interpolate</span><span class="p">(</span><span class="n">method</span><span class="o">=</span><span class="s1">'values'</span><span class="p">)</span>
<span class="gr">Out[64]: </span>
<span class="go">0.0 0.0</span>
<span class="go">1.0 1.0</span>
<span class="go">10.0 10.0</span>
<span class="go">dtype: float64</span>
</pre></div>
</div>
<p><span class="yiyi-st" id="yiyi-149">您还可以使用DataFrame插值：</span></p>
<div class="highlight-ipython"><div class="highlight"><pre><span></span><span class="gp">In [65]: </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="s1">'A'</span><span class="p">:</span> <span class="p">[</span><span class="mi">1</span><span class="p">,</span> <span class="mf">2.1</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="mf">4.7</span><span class="p">,</span> <span class="mf">5.6</span><span class="p">,</span> <span class="mf">6.8</span><span class="p">],</span>
<span class="gp"> ....:</span> <span class="s1">'B'</span><span class="p">:</span> <span class="p">[</span><span class="o">.</span><span class="mi">25</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="n">np</span><span class="o">.</span><span class="n">nan</span><span class="p">,</span> <span class="mi">4</span><span class="p">,</span> <span class="mf">12.2</span><span class="p">,</span> <span class="mf">14.4</span><span class="p">]})</span>
<span class="gp"> ....:</span>
<span class="gp">In [66]: </span><span class="n">df</span>
<span class="gr">Out[66]: </span>
<span class="go"> A B</span>
<span class="go">0 1.0 0.25</span>
<span class="go">1 2.1 NaN</span>
<span class="go">2 NaN NaN</span>
<span class="go">3 4.7 4.00</span>
<span class="go">4 5.6 12.20</span>
<span class="go">5 6.8 14.40</span>
<span class="gp">In [67]: </span><span class="n">df</span><span class="o">.</span><span class="n">interpolate</span><span class="p">()</span>
<span class="gr">Out[67]: </span>
<span class="go"> A B</span>
<span class="go">0 1.0 0.25</span>
<span class="go">1 2.1 1.50</span>
<span class="go">2 3.4 2.75</span>
<span class="go">3 4.7 4.00</span>
<span class="go">4 5.6 12.20</span>
<span class="go">5 6.8 14.40</span>
</pre></div>
</div>
<p><span class="yiyi-st" id="yiyi-150"><code class="docutils literal"><span class="pre">method</span></code>参数允许访问fancier插值方法。</span><span class="yiyi-st" id="yiyi-151">如果您安装了<a class="reference external" href="http://www.scipy.org">scipy</a>，则可以将1-d插值程序的名称设置为<code class="docutils literal"><span class="pre">method</span></code>。</span><span class="yiyi-st" id="yiyi-152">有关详细信息，请参阅完整的scipy插值<a class="reference external" href="http://docs.scipy.org/doc/scipy/reference/interpolate.html#univariate-interpolation">文档</a>和参考<a class="reference external" href="http://docs.scipy.org/doc/scipy/reference/tutorial/interpolate.html">指南</a>。</span><span class="yiyi-st" id="yiyi-153">适当的插值方法将取决于您使用的数据类型。</span></p>
<ul class="simple">
<li><span class="yiyi-st" id="yiyi-154">如果你正在处理一个以递增的速度增长的时间序列，则<code class="docutils literal"><span class="pre">method='quadratic'</span></code>可能是适当的。</span></li>
<li><span class="yiyi-st" id="yiyi-155">如果你有近似累积分布函数的值，那么<code class="docutils literal"><span class="pre">method='pchip'</span></code>应该工作得很好。</span></li>
<li><span class="yiyi-st" id="yiyi-156">要以平滑绘图的目标填充缺失值，请使用<code class="docutils literal"><span class="pre">method='akima'</span></code>。</span></li>
</ul>
<div class="admonition warning">
<p class="first admonition-title"><span class="yiyi-st" id="yiyi-157">警告</span></p>
<p class="last"><span class="yiyi-st" id="yiyi-158">这些方法需要<code class="docutils literal"><span class="pre">scipy</span></code>。</span></p>
</div>
<div class="highlight-ipython"><div class="highlight"><pre><span></span><span class="gp">In [68]: </span><span class="n">df</span><span class="o">.</span><span class="n">interpolate</span><span class="p">(</span><span class="n">method</span><span class="o">=</span><span class="s1">'barycentric'</span><span class="p">)</span>
<span class="gr">Out[68]: </span>
<span class="go"> A B</span>
<span class="go">0 1.00 0.250</span>
<span class="go">1 2.10 -7.660</span>
<span class="go">2 3.53 -4.515</span>
<span class="go">3 4.70 4.000</span>
<span class="go">4 5.60 12.200</span>
<span class="go">5 6.80 14.400</span>
<span class="gp">In [69]: </span><span class="n">df</span><span class="o">.</span><span class="n">interpolate</span><span class="p">(</span><span class="n">method</span><span class="o">=</span><span class="s1">'pchip'</span><span class="p">)</span>
<span class="gr">Out[69]: </span>
<span class="go"> A B</span>
<span class="go">0 1.00000 0.250000</span>
<span class="go">1 2.10000 0.672808</span>
<span class="go">2 3.43454 1.928950</span>
<span class="go">3 4.70000 4.000000</span>
<span class="go">4 5.60000 12.200000</span>
<span class="go">5 6.80000 14.400000</span>
<span class="gp">In [70]: </span><span class="n">df</span><span class="o">.</span><span class="n">interpolate</span><span class="p">(</span><span class="n">method</span><span class="o">=</span><span class="s1">'akima'</span><span class="p">)</span>
<span class="gr">Out[70]: </span>
<span class="go"> A B</span>
<span class="go">0 1.000000 0.250000</span>
<span class="go">1 2.100000 -0.873316</span>
<span class="go">2 3.406667 0.320034</span>
<span class="go">3 4.700000 4.000000</span>
<span class="go">4 5.600000 12.200000</span>
<span class="go">5 6.800000 14.400000</span>
</pre></div>
</div>
<p><span class="yiyi-st" id="yiyi-159">当通过多项式或样条逼近进行插值时，还必须指定近似的次数或次数：</span></p>
<div class="highlight-ipython"><div class="highlight"><pre><span></span><span class="gp">In [71]: </span><span class="n">df</span><span class="o">.</span><span class="n">interpolate</span><span class="p">(</span><span class="n">method</span><span class="o">=</span><span class="s1">'spline'</span><span class="p">,</span> <span class="n">order</span><span class="o">=</span><span class="mi">2</span><span class="p">)</span>
<span class="gr">Out[71]: </span>
<span class="go"> A B</span>
<span class="go">0 1.000000 0.250000</span>
<span class="go">1 2.100000 -0.428598</span>
<span class="go">2 3.404545 1.206900</span>
<span class="go">3 4.700000 4.000000</span>
<span class="go">4 5.600000 12.200000</span>
<span class="go">5 6.800000 14.400000</span>
<span class="gp">In [72]: </span><span class="n">df</span><span class="o">.</span><span class="n">interpolate</span><span class="p">(</span><span class="n">method</span><span class="o">=</span><span class="s1">'polynomial'</span><span class="p">,</span> <span class="n">order</span><span class="o">=</span><span class="mi">2</span><span class="p">)</span>
<span class="gr">Out[72]: </span>
<span class="go"> A B</span>
<span class="go">0 1.000000 0.250000</span>
<span class="go">1 2.100000 -4.161538</span>
<span class="go">2 3.547059 -2.911538</span>
<span class="go">3 4.700000 4.000000</span>
<span class="go">4 5.600000 12.200000</span>
<span class="go">5 6.800000 14.400000</span>
</pre></div>
</div>
<p><span class="yiyi-st" id="yiyi-160">比较几种方法：</span></p>
<div class="highlight-ipython"><div class="highlight"><pre><span></span><span class="gp">In [73]: </span><span class="n">np</span><span class="o">.</span><span class="n">random</span><span class="o">.</span><span class="n">seed</span><span class="p">(</span><span class="mi">2</span><span class="p">)</span>
<span class="gp">In [74]: </span><span class="n">ser</span> <span class="o">=</span> <span class="n">pd</span><span class="o">.</span><span class="n">Series</span><span class="p">(</span><span class="n">np</span><span class="o">.</span><span class="n">arange</span><span class="p">(</span><span class="mi">1</span><span class="p">,</span> <span class="mf">10.1</span><span class="p">,</span> <span class="o">.</span><span class="mi">25</span><span class="p">)</span><span class="o">**</span><span class="mi">2</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">randn</span><span class="p">(</span><span class="mi">37</span><span class="p">))</span>
<span class="gp">In [75]: </span><span class="n">bad</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="mi">4</span><span class="p">,</span> <span class="mi">13</span><span class="p">,</span> <span class="mi">14</span><span class="p">,</span> <span class="mi">15</span><span class="p">,</span> <span class="mi">16</span><span class="p">,</span> <span class="mi">17</span><span class="p">,</span> <span class="mi">18</span><span class="p">,</span> <span class="mi">20</span><span class="p">,</span> <span class="mi">29</span><span class="p">])</span>
<span class="gp">In [76]: </span><span class="n">ser</span><span class="p">[</span><span class="n">bad</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">In [77]: </span><span class="n">methods</span> <span class="o">=</span> <span class="p">[</span><span class="s1">'linear'</span><span class="p">,</span> <span class="s1">'quadratic'</span><span class="p">,</span> <span class="s1">'cubic'</span><span class="p">]</span>
<span class="gp">In [78]: </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="n">m</span><span class="p">:</span> <span class="n">ser</span><span class="o">.</span><span class="n">interpolate</span><span class="p">(</span><span class="n">method</span><span class="o">=</span><span class="n">m</span><span class="p">)</span> <span class="k">for</span> <span class="n">m</span> <span class="ow">in</span> <span class="n">methods</span><span class="p">})</span>
<span class="gp">In [79]: </span><span class="n">df</span><span class="o">.</span><span class="n">plot</span><span class="p">()</span>
<span class="gr">Out[79]: </span><span class="o"><</span><span class="n">matplotlib</span><span class="o">.</span><span class="n">axes</span><span class="o">.</span><span class="n">_subplots</span><span class="o">.</span><span class="n">AxesSubplot</span> <span class="n">at</span> <span class="mh">0x7ff2666771d0</span><span class="o">></span>
</pre></div>
</div>
<img alt="http://pandas.pydata.org/pandas-docs/version/0.19.2/_images/compare_interpolations.png" src="http://pandas.pydata.org/pandas-docs/version/0.19.2/_images/compare_interpolations.png">
<p><span class="yiyi-st" id="yiyi-161">另一个用例是在<em>新</em>值处的插值。</span><span class="yiyi-st" id="yiyi-162">假设你有一些分布的100个观察。</span><span class="yiyi-st" id="yiyi-163">让我们假设你对中间发生的事情特别感兴趣。</span><span class="yiyi-st" id="yiyi-164">您可以混合使用pandas'<code class="docutils literal"><span class="pre">reindex</span></code>和<code class="docutils literal"><span class="pre">interpolate</span></code>方法在新值处插值。</span></p>
<div class="highlight-ipython"><div class="highlight"><pre><span></span><span class="gp">In [80]: </span><span class="n">ser</span> <span class="o">=</span> <span class="n">pd</span><span class="o">.</span><span class="n">Series</span><span class="p">(</span><span class="n">np</span><span class="o">.</span><span class="n">sort</span><span class="p">(</span><span class="n">np</span><span class="o">.</span><span class="n">random</span><span class="o">.</span><span class="n">uniform</span><span class="p">(</span><span class="n">size</span><span class="o">=</span><span class="mi">100</span><span class="p">)))</span>
<span class="c"># interpolate at new_index</span>
<span class="gp">In [81]: </span><span class="n">new_index</span> <span class="o">=</span> <span class="n">ser</span><span class="o">.</span><span class="n">index</span> <span class="o">|</span> <span class="n">pd</span><span class="o">.</span><span class="n">Index</span><span class="p">([</span><span class="mf">49.25</span><span class="p">,</span> <span class="mf">49.5</span><span class="p">,</span> <span class="mf">49.75</span><span class="p">,</span> <span class="mf">50.25</span><span class="p">,</span> <span class="mf">50.5</span><span class="p">,</span> <span class="mf">50.75</span><span class="p">])</span>
<span class="gp">In [82]: </span><span class="n">interp_s</span> <span class="o">=</span> <span class="n">ser</span><span class="o">.</span><span class="n">reindex</span><span class="p">(</span><span class="n">new_index</span><span class="p">)</span><span class="o">.</span><span class="n">interpolate</span><span class="p">(</span><span class="n">method</span><span class="o">=</span><span class="s1">'pchip'</span><span class="p">)</span>
<span class="gp">In [83]: </span><span class="n">interp_s</span><span class="p">[</span><span class="mi">49</span><span class="p">:</span><span class="mi">51</span><span class="p">]</span>
<span class="gr">Out[83]: </span>
<span class="go">49.00 0.471410</span>
<span class="go">49.25 0.476841</span>
<span class="go">49.50 0.481780</span>
<span class="go">49.75 0.485998</span>
<span class="go">50.00 0.489266</span>
<span class="go">50.25 0.491814</span>
<span class="go">50.50 0.493995</span>
<span class="go">50.75 0.495763</span>
<span class="go">51.00 0.497074</span>
<span class="go">dtype: float64</span>
</pre></div>
</div>
<div class="section" id="interpolation-limits">
<h4><span class="yiyi-st" id="yiyi-165">Interpolation Limits</span></h4>
<p><span class="yiyi-st" id="yiyi-166">与其他Pandas填充方法一样，<code class="docutils literal"><span class="pre">interpolate</span></code>接受一个<code class="docutils literal"><span class="pre">limit</span></code>关键字参数。</span><span class="yiyi-st" id="yiyi-167">使用此参数限制连续插值的数量，对于距离上次有效观测值过远的插值，保留<code class="docutils literal"><span class="pre">NaN</span></code>值：</span></p>
<div class="highlight-ipython"><div class="highlight"><pre><span></span><span class="gp">In [84]: </span><span class="n">ser</span> <span class="o">=</span> <span class="n">pd</span><span class="o">.</span><span class="n">Series</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="n">np</span><span class="o">.</span><span class="n">nan</span><span class="p">,</span> <span class="mi">5</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="n">np</span><span class="o">.</span><span class="n">nan</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">13</span><span class="p">])</span>
<span class="gp">In [85]: </span><span class="n">ser</span><span class="o">.</span><span class="n">interpolate</span><span class="p">(</span><span class="n">limit</span><span class="o">=</span><span class="mi">2</span><span class="p">)</span>
<span class="gr">Out[85]: </span>
<span class="go">0 NaN</span>
<span class="go">1 NaN</span>
<span class="go">2 5.0</span>
<span class="go">3 7.0</span>
<span class="go">4 9.0</span>
<span class="go">5 NaN</span>
<span class="go">6 13.0</span>
<span class="go">dtype: float64</span>
</pre></div>
</div>
<p><span class="yiyi-st" id="yiyi-168">默认情况下，<code class="docutils literal"><span class="pre">limit</span></code>适用于正向，因此在非<code class="docutils literal"><span class="pre">NaN</span></code>值后只能填充<code class="docutils literal"><span class="pre">NaN</span></code>值。</span><span class="yiyi-st" id="yiyi-169">If you provide <code class="docutils literal"><span class="pre">'backward'</span></code> or <code class="docutils literal"><span class="pre">'both'</span></code> for the <code class="docutils literal"><span class="pre">limit_direction</span></code> keyword argument, you can fill <code class="docutils literal"><span class="pre">NaN</span></code> values before non-<code class="docutils literal"><span class="pre">NaN</span></code> values, or both before and after non-<code class="docutils literal"><span class="pre">NaN</span></code> values, respectively:</span></p>
<div class="highlight-ipython"><div class="highlight"><pre><span></span><span class="gp">In [86]: </span><span class="n">ser</span><span class="o">.</span><span class="n">interpolate</span><span class="p">(</span><span class="n">limit</span><span class="o">=</span><span class="mi">1</span><span class="p">)</span> <span class="c1"># limit_direction == 'forward'</span>
<span class="gr">Out[86]: </span>
<span class="go">0 NaN</span>
<span class="go">1 NaN</span>
<span class="go">2 5.0</span>
<span class="go">3 7.0</span>
<span class="go">4 NaN</span>
<span class="go">5 NaN</span>
<span class="go">6 13.0</span>
<span class="go">dtype: float64</span>
<span class="gp">In [87]: </span><span class="n">ser</span><span class="o">.</span><span class="n">interpolate</span><span class="p">(</span><span class="n">limit</span><span class="o">=</span><span class="mi">1</span><span class="p">,</span> <span class="n">limit_direction</span><span class="o">=</span><span class="s1">'backward'</span><span class="p">)</span>
<span class="gr">Out[87]: </span>
<span class="go">0 NaN</span>
<span class="go">1 5.0</span>
<span class="go">2 5.0</span>
<span class="go">3 NaN</span>
<span class="go">4 NaN</span>
<span class="go">5 11.0</span>
<span class="go">6 13.0</span>
<span class="go">dtype: float64</span>
<span class="gp">In [88]: </span><span class="n">ser</span><span class="o">.</span><span class="n">interpolate</span><span class="p">(</span><span class="n">limit</span><span class="o">=</span><span class="mi">1</span><span class="p">,</span> <span class="n">limit_direction</span><span class="o">=</span><span class="s1">'both'</span><span class="p">)</span>
<span class="gr">Out[88]: </span>
<span class="go">0 NaN</span>
<span class="go">1 5.0</span>
<span class="go">2 5.0</span>
<span class="go">3 7.0</span>
<span class="go">4 NaN</span>
<span class="go">5 11.0</span>
<span class="go">6 13.0</span>
<span class="go">dtype: float64</span>
</pre></div>
</div>
</div>
</div>
<div class="section" id="replacing-generic-values">
<span id="missing-data-replace"></span><h3><span class="yiyi-st" id="yiyi-170">Replacing Generic Values</span></h3>
<p><span class="yiyi-st" id="yiyi-171">通常我们想用其他值替换任意值。</span><span class="yiyi-st" id="yiyi-172">v0.8中的新增功能是Series / DataFrame中的<code class="docutils literal"><span class="pre">replace</span></code>方法，它提供了一种高效而灵活的方法来执行此类替换。</span></p>
<p><span class="yiyi-st" id="yiyi-173">对于系列，可以使用另一个值替换单个值或值列表：</span></p>
<div class="highlight-ipython"><div class="highlight"><pre><span></span><span class="gp">In [89]: </span><span class="n">ser</span> <span class="o">=</span> <span class="n">pd</span><span class="o">.</span><span class="n">Series</span><span class="p">([</span><span class="mf">0.</span><span class="p">,</span> <span class="mf">1.</span><span class="p">,</span> <span class="mf">2.</span><span class="p">,</span> <span class="mf">3.</span><span class="p">,</span> <span class="mf">4.</span><span class="p">])</span>
<span class="gp">In [90]: </span><span class="n">ser</span><span class="o">.</span><span class="n">replace</span><span class="p">(</span><span class="mi">0</span><span class="p">,</span> <span class="mi">5</span><span class="p">)</span>
<span class="gr">Out[90]: </span>
<span class="go">0 5.0</span>
<span class="go">1 1.0</span>
<span class="go">2 2.0</span>
<span class="go">3 3.0</span>
<span class="go">4 4.0</span>
<span class="go">dtype: float64</span>
</pre></div>
</div>
<p><span class="yiyi-st" id="yiyi-174">您可以使用其他值列表替换值列表：</span></p>
<div class="highlight-ipython"><div class="highlight"><pre><span></span><span class="gp">In [91]: </span><span class="n">ser</span><span class="o">.</span><span class="n">replace</span><span class="p">([</span><span class="mi">0</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="mi">3</span><span class="p">,</span> <span class="mi">4</span><span class="p">],</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="mi">2</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="gr">Out[91]: </span>
<span class="go">0 4.0</span>
<span class="go">1 3.0</span>
<span class="go">2 2.0</span>
<span class="go">3 1.0</span>
<span class="go">4 0.0</span>
<span class="go">dtype: float64</span>
</pre></div>
</div>
<p><span class="yiyi-st" id="yiyi-175">您还可以指定映射dict：</span></p>
<div class="highlight-ipython"><div class="highlight"><pre><span></span><span class="gp">In [92]: </span><span class="n">ser</span><span class="o">.</span><span class="n">replace</span><span class="p">({</span><span class="mi">0</span><span class="p">:</span> <span class="mi">10</span><span class="p">,</span> <span class="mi">1</span><span class="p">:</span> <span class="mi">100</span><span class="p">})</span>
<span class="gr">Out[92]: </span>
<span class="go">0 10.0</span>
<span class="go">1 100.0</span>
<span class="go">2 2.0</span>
<span class="go">3 3.0</span>
<span class="go">4 4.0</span>
<span class="go">dtype: float64</span>
</pre></div>
</div>
<p><span class="yiyi-st" id="yiyi-176">对于DataFrame，您可以按列指定单个值：</span></p>
<div class="highlight-ipython"><div class="highlight"><pre><span></span><span class="gp">In [93]: </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="s1">'a'</span><span class="p">:</span> <span class="p">[</span><span class="mi">0</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="mi">3</span><span class="p">,</span> <span class="mi">4</span><span class="p">],</span> <span class="s1">'b'</span><span class="p">:</span> <span class="p">[</span><span class="mi">5</span><span class="p">,</span> <span class="mi">6</span><span class="p">,</span> <span class="mi">7</span><span class="p">,</span> <span class="mi">8</span><span class="p">,</span> <span class="mi">9</span><span class="p">]})</span>
<span class="gp">In [94]: </span><span class="n">df</span><span class="o">.</span><span class="n">replace</span><span class="p">({</span><span class="s1">'a'</span><span class="p">:</span> <span class="mi">0</span><span class="p">,</span> <span class="s1">'b'</span><span class="p">:</span> <span class="mi">5</span><span class="p">},</span> <span class="mi">100</span><span class="p">)</span>
<span class="gr">Out[94]: </span>
<span class="go"> a b</span>
<span class="go">0 100 100</span>
<span class="go">1 1 6</span>
<span class="go">2 2 7</span>
<span class="go">3 3 8</span>
<span class="go">4 4 9</span>
</pre></div>
</div>
<p><span class="yiyi-st" id="yiyi-177">您可以将所有给定的值视为缺失并对其进行插值，而不是替换为指定的值：</span></p>
<div class="highlight-ipython"><div class="highlight"><pre><span></span><span class="gp">In [95]: </span><span class="n">ser</span><span class="o">.</span><span class="n">replace</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="mi">3</span><span class="p">],</span> <span class="n">method</span><span class="o">=</span><span class="s1">'pad'</span><span class="p">)</span>
<span class="gr">Out[95]: </span>
<span class="go">0 0.0</span>
<span class="go">1 0.0</span>
<span class="go">2 0.0</span>
<span class="go">3 0.0</span>
<span class="go">4 4.0</span>
<span class="go">dtype: float64</span>
</pre></div>
</div>
</div>
<div class="section" id="string-regular-expression-replacement">
<span id="missing-data-replace-expression"></span><h3><span class="yiyi-st" id="yiyi-178">String/Regular Expression Replacement</span></h3>
<div class="admonition note">
<p class="first admonition-title"><span class="yiyi-st" id="yiyi-179">注意</span></p>
<p class="last"><span class="yiyi-st" id="yiyi-180">以<code class="docutils literal"><span class="pre">r</span></code>字符为前缀的Python字符串，如<code class="docutils literal"><span class="pre">r'hello</span> <span class="pre">world'</span></code>是所谓的“raw”字符串。</span><span class="yiyi-st" id="yiyi-181">他们有不同的反斜杠语义比没有这个前缀的字符串。</span><span class="yiyi-st" id="yiyi-182">原始字符串中的反斜杠将被解释为转义的反斜杠，例如<code class="docutils literal"><span class="pre">r'\'</span> <span class="pre">==</span> <span class="pre">'\\' t0>。</span></code></span><span class="yiyi-st" id="yiyi-183">如果这不清楚，您应该<a class="reference external" href="http://docs.python.org/2/reference/lexical_analysis.html#string-literals">阅读他们</a>。</span></p>
</div>
<p><span class="yiyi-st" id="yiyi-184">用<code class="docutils literal"><span class="pre">nan</span></code>（str - > str）替换'。'</span></p>
<div class="highlight-ipython"><div class="highlight"><pre><span></span><span class="gp">In [96]: </span><span class="n">d</span> <span class="o">=</span> <span class="p">{</span><span class="s1">'a'</span><span class="p">:</span> <span class="nb">list</span><span class="p">(</span><span class="nb">range</span><span class="p">(</span><span class="mi">4</span><span class="p">)),</span> <span class="s1">'b'</span><span class="p">:</span> <span class="nb">list</span><span class="p">(</span><span class="s1">'ab..'</span><span class="p">),</span> <span class="s1">'c'</span><span class="p">:</span> <span class="p">[</span><span class="s1">'a'</span><span class="p">,</span> <span class="s1">'b'</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="s1">'d'</span><span class="p">]}</span>
<span class="gp">In [97]: </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="n">d</span><span class="p">)</span>
<span class="gp">In [98]: </span><span class="n">df</span><span class="o">.</span><span class="n">replace</span><span class="p">(</span><span class="s1">'.'</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="gr">Out[98]: </span>
<span class="go"> a b c</span>
<span class="go">0 0 a a</span>
<span class="go">1 1 b b</span>
<span class="go">2 2 NaN NaN</span>
<span class="go">3 3 NaN d</span>
</pre></div>
</div>
<p><span class="yiyi-st" id="yiyi-185">现在使用正则表达式删除周围的空格（regex - > regex）</span></p>
<div class="highlight-ipython"><div class="highlight"><pre><span></span><span class="gp">In [99]: </span><span class="n">df</span><span class="o">.</span><span class="n">replace</span><span class="p">(</span><span class="s1">r'\s*\.\s*'</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="n">regex</span><span class="o">=</span><span class="bp">True</span><span class="p">)</span>
<span class="gr">Out[99]: </span>
<span class="go"> a b c</span>
<span class="go">0 0 a a</span>
<span class="go">1 1 b b</span>
<span class="go">2 2 NaN NaN</span>
<span class="go">3 3 NaN d</span>
</pre></div>
</div>
<p><span class="yiyi-st" id="yiyi-186">替换几个不同的值（list - > list）</span></p>
<div class="highlight-ipython"><div class="highlight"><pre><span></span><span class="gp">In [100]: </span><span class="n">df</span><span class="o">.</span><span class="n">replace</span><span class="p">([</span><span class="s1">'a'</span><span class="p">,</span> <span class="s1">'.'</span><span class="p">],</span> <span class="p">[</span><span class="s1">'b'</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="gr">Out[100]: </span>
<span class="go"> a b c</span>
<span class="go">0 0 b b</span>
<span class="go">1 1 b b</span>
<span class="go">2 2 NaN NaN</span>
<span class="go">3 3 NaN d</span>
</pre></div>
</div>
<p><span class="yiyi-st" id="yiyi-187">regex列表 - > regex列表</span></p>
<div class="highlight-ipython"><div class="highlight"><pre><span></span><span class="gp">In [101]: </span><span class="n">df</span><span class="o">.</span><span class="n">replace</span><span class="p">([</span><span class="s1">r'\.'</span><span class="p">,</span> <span class="s1">r'(a)'</span><span class="p">],</span> <span class="p">[</span><span class="s1">'dot'</span><span class="p">,</span> <span class="s1">'</span><span class="se">\1</span><span class="s1">stuff'</span><span class="p">],</span> <span class="n">regex</span><span class="o">=</span><span class="bp">True</span><span class="p">)</span>
<span class="gr">Out[101]: </span>
<span class="go"> a b c</span>
<span class="go">0 0 stuff stuff</span>
<span class="go">1 1 b b</span>
<span class="go">2 2 dot NaN</span>
<span class="go">3 3 dot d</span>
</pre></div>
</div>
<p><span class="yiyi-st" id="yiyi-188">只搜索列<code class="docutils literal"><span class="pre">'b'</span></code>（dict - > dict）</span></p>
<div class="highlight-ipython"><div class="highlight"><pre><span></span><span class="gp">In [102]: </span><span class="n">df</span><span class="o">.</span><span class="n">replace</span><span class="p">({</span><span class="s1">'b'</span><span class="p">:</span> <span class="s1">'.'</span><span class="p">},</span> <span class="p">{</span><span class="s1">'b'</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="gr">Out[102]: </span>
<span class="go"> a b c</span>
<span class="go">0 0 a a</span>
<span class="go">1 1 b b</span>
<span class="go">2 2 NaN NaN</span>
<span class="go">3 3 NaN d</span>
</pre></div>
</div>
<p><span class="yiyi-st" id="yiyi-189">与上一个示例相同，但使用正则表达式进行搜索（dict的regex - > dict）</span></p>
<div class="highlight-ipython"><div class="highlight"><pre><span></span><span class="gp">In [103]: </span><span class="n">df</span><span class="o">.</span><span class="n">replace</span><span class="p">({</span><span class="s1">'b'</span><span class="p">:</span> <span class="s1">r'\s*\.\s*'</span><span class="p">},</span> <span class="p">{</span><span class="s1">'b'</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="n">regex</span><span class="o">=</span><span class="bp">True</span><span class="p">)</span>
<span class="gr">Out[103]: </span>
<span class="go"> a b c</span>
<span class="go">0 0 a a</span>
<span class="go">1 1 b b</span>
<span class="go">2 2 NaN NaN</span>
<span class="go">3 3 NaN d</span>
</pre></div>
</div>
<p><span class="yiyi-st" id="yiyi-190">您可以传递使用<code class="docutils literal"><span class="pre">regex=True</span></code>的正则表达式的嵌套字典</span></p>
<div class="highlight-ipython"><div class="highlight"><pre><span></span><span class="gp">In [104]: </span><span class="n">df</span><span class="o">.</span><span class="n">replace</span><span class="p">({</span><span class="s1">'b'</span><span class="p">:</span> <span class="p">{</span><span class="s1">'b'</span><span class="p">:</span> <span class="s1">r''</span><span class="p">}},</span> <span class="n">regex</span><span class="o">=</span><span class="bp">True</span><span class="p">)</span>
<span class="gr">Out[104]: </span>
<span class="go"> a b c</span>
<span class="go">0 0 a a</span>
<span class="go">1 1 b</span>
<span class="go">2 2 . NaN</span>
<span class="go">3 3 . d</span>
</pre></div>
</div>
<p><span class="yiyi-st" id="yiyi-191">或者你可以像这样传递嵌套字典</span></p>
<div class="highlight-ipython"><div class="highlight"><pre><span></span><span class="gp">In [105]: </span><span class="n">df</span><span class="o">.</span><span class="n">replace</span><span class="p">(</span><span class="n">regex</span><span class="o">=</span><span class="p">{</span><span class="s1">'b'</span><span class="p">:</span> <span class="p">{</span><span class="s1">r'\s*\.\s*'</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="gr">Out[105]: </span>
<span class="go"> a b c</span>
<span class="go">0 0 a a</span>
<span class="go">1 1 b b</span>
<span class="go">2 2 NaN NaN</span>
<span class="go">3 3 NaN d</span>
</pre></div>
</div>
<p><span class="yiyi-st" id="yiyi-192">你也可以在替换时使用正则表达式匹配的组（正则表达式的dict - >正则表达式的dict），这也适用于列表</span></p>
<div class="highlight-ipython"><div class="highlight"><pre><span></span><span class="gp">In [106]: </span><span class="n">df</span><span class="o">.</span><span class="n">replace</span><span class="p">({</span><span class="s1">'b'</span><span class="p">:</span> <span class="s1">r'\s*(\.)\s*'</span><span class="p">},</span> <span class="p">{</span><span class="s1">'b'</span><span class="p">:</span> <span class="s1">r'\1ty'</span><span class="p">},</span> <span class="n">regex</span><span class="o">=</span><span class="bp">True</span><span class="p">)</span>
<span class="gr">Out[106]: </span>
<span class="go"> a b c</span>
<span class="go">0 0 a a</span>
<span class="go">1 1 b b</span>
<span class="go">2 2 .ty NaN</span>
<span class="go">3 3 .ty d</span>
</pre></div>
</div>
<p><span class="yiyi-st" id="yiyi-193">您可以传递正则表达式列表，其中匹配的那些将被替换为标量（regex - > regex列表）</span></p>
<div class="highlight-ipython"><div class="highlight"><pre><span></span><span class="gp">In [107]: </span><span class="n">df</span><span class="o">.</span><span class="n">replace</span><span class="p">([</span><span class="s1">r'\s*\.\s*'</span><span class="p">,</span> <span class="s1">r'a|b'</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="n">regex</span><span class="o">=</span><span class="bp">True</span><span class="p">)</span>
<span class="gr">Out[107]: </span>
<span class="go"> a b c</span>
<span class="go">0 0 NaN NaN</span>
<span class="go">1 1 NaN NaN</span>
<span class="go">2 2 NaN NaN</span>
<span class="go">3 3 NaN d</span>
</pre></div>
</div>
<p><span class="yiyi-st" id="yiyi-194">所有正则表达式示例也可以使用<code class="docutils literal"><span class="pre">to_replace</span></code>参数作为<code class="docutils literal"><span class="pre">regex</span></code>参数传递。</span><span class="yiyi-st" id="yiyi-195">在这种情况下，<code class="docutils literal"><span class="pre">value</span></code>参数必须通过名称显式传递，或<code class="docutils literal"><span class="pre">regex</span></code>必须是嵌套字典。</span><span class="yiyi-st" id="yiyi-196">在这种情况下，上一个示例将是</span></p>
<div class="highlight-ipython"><div class="highlight"><pre><span></span><span class="gp">In [108]: </span><span class="n">df</span><span class="o">.</span><span class="n">replace</span><span class="p">(</span><span class="n">regex</span><span class="o">=</span><span class="p">[</span><span class="s1">r'\s*\.\s*'</span><span class="p">,</span> <span class="s1">r'a|b'</span><span class="p">],</span> <span class="n">value</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="gr">Out[108]: </span>
<span class="go"> a b c</span>
<span class="go">0 0 NaN NaN</span>
<span class="go">1 1 NaN NaN</span>
<span class="go">2 2 NaN NaN</span>
<span class="go">3 3 NaN d</span>
</pre></div>
</div>
<p><span class="yiyi-st" id="yiyi-197">如果您不想在每次要使用正则表达式时传递<code class="docutils literal"><span class="pre">regex=True</span></code>，这都很方便。</span></p>
<div class="admonition note">
<p class="first admonition-title"><span class="yiyi-st" id="yiyi-198">注意</span></p>
<p class="last"><span class="yiyi-st" id="yiyi-199">在上述<code class="docutils literal"><span class="pre">replace</span></code>示例的任何地方，您看到正则表达式编译的正则表达式也是有效的。</span></p>
</div>
</div>
<div class="section" id="numeric-replacement">
<h3><span class="yiyi-st" id="yiyi-200">Numeric Replacement</span></h3>
<p><span class="yiyi-st" id="yiyi-201">类似于<code class="docutils literal"><span class="pre">DataFrame.fillna</span></code></span></p>
<div class="highlight-ipython"><div class="highlight"><pre><span></span><span class="gp">In [109]: </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="n">np</span><span class="o">.</span><span class="n">random</span><span class="o">.</span><span class="n">randn</span><span class="p">(</span><span class="mi">10</span><span class="p">,</span> <span class="mi">2</span><span class="p">))</span>
<span class="gp">In [110]: </span><span class="n">df</span><span class="p">[</span><span class="n">np</span><span class="o">.</span><span class="n">random</span><span class="o">.</span><span class="n">rand</span><span class="p">(</span><span class="n">df</span><span class="o">.</span><span class="n">shape</span><span class="p">[</span><span class="mi">0</span><span class="p">])</span> <span class="o">></span> <span class="mf">0.5</span><span class="p">]</span> <span class="o">=</span> <span class="mf">1.5</span>
<span class="gp">In [111]: </span><span class="n">df</span><span class="o">.</span><span class="n">replace</span><span class="p">(</span><span class="mf">1.5</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="gr">Out[111]: </span>
<span class="go"> 0 1</span>
<span class="go">0 -0.844214 -1.021415</span>
<span class="go">1 0.432396 -0.323580</span>
<span class="go">2 0.423825 0.799180</span>
<span class="go">3 1.262614 0.751965</span>
<span class="go">4 NaN NaN</span>
<span class="go">5 NaN NaN</span>
<span class="go">6 -0.498174 -1.060799</span>
<span class="go">7 0.591667 -0.183257</span>
<span class="go">8 1.019855 -1.482465</span>
<span class="go">9 NaN NaN</span>
</pre></div>
</div>
<p><span class="yiyi-st" id="yiyi-202">通过列表替换多个值也是如此</span></p>
<div class="highlight-ipython"><div class="highlight"><pre><span></span><span class="gp">In [112]: </span><span class="n">df00</span> <span class="o">=</span> <span class="n">df</span><span class="o">.</span><span class="n">values</span><span class="p">[</span><span class="mi">0</span><span class="p">,</span> <span class="mi">0</span><span class="p">]</span>
<span class="gp">In [113]: </span><span class="n">df</span><span class="o">.</span><span class="n">replace</span><span class="p">([</span><span class="mf">1.5</span><span class="p">,</span> <span class="n">df00</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="s1">'a'</span><span class="p">])</span>
<span class="gr">Out[113]: </span>
<span class="go"> 0 1</span>
<span class="go">0 a -1.021415</span>