-
Notifications
You must be signed in to change notification settings - Fork 31
/
PermutationFeatureImportanceExtensions.xml
573 lines (573 loc) · 57.1 KB
/
PermutationFeatureImportanceExtensions.xml
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
<Type Name="PermutationFeatureImportanceExtensions" FullName="Microsoft.ML.PermutationFeatureImportanceExtensions">
<TypeSignature Language="C#" Value="public static class PermutationFeatureImportanceExtensions" />
<TypeSignature Language="ILAsm" Value=".class public auto ansi abstract sealed beforefieldinit PermutationFeatureImportanceExtensions extends System.Object" />
<TypeSignature Language="DocId" Value="T:Microsoft.ML.PermutationFeatureImportanceExtensions" />
<TypeSignature Language="VB.NET" Value="Public Module PermutationFeatureImportanceExtensions" />
<TypeSignature Language="F#" Value="type PermutationFeatureImportanceExtensions = class" />
<AssemblyInfo>
<AssemblyName>Microsoft.ML.Transforms</AssemblyName>
<AssemblyVersion>1.0.0.0</AssemblyVersion>
</AssemblyInfo>
<Base>
<BaseTypeName>System.Object</BaseTypeName>
</Base>
<Interfaces />
<Docs>
<summary>
Collection of extension methods used by <see cref="T:Microsoft.ML.RegressionCatalog" />,
<see cref="T:Microsoft.ML.BinaryClassificationCatalog" />, <see cref="T:Microsoft.ML.MulticlassClassificationCatalog" />,
and <see cref="T:Microsoft.ML.RankingCatalog" /> to create instances of permutation feature importance components.
</summary>
<remarks>To be added.</remarks>
</Docs>
<Members>
<Member MemberName="PermutationFeatureImportance">
<MemberSignature Language="C#" Value="public static System.Collections.Immutable.ImmutableDictionary<string,Microsoft.ML.Data.MulticlassClassificationMetricsStatistics> PermutationFeatureImportance (this Microsoft.ML.MulticlassClassificationCatalog catalog, Microsoft.ML.ITransformer model, Microsoft.ML.IDataView data, string labelColumnName = "Label", bool useFeatureWeightFilter = false, int? numberOfExamplesToUse = default, int permutationCount = 1);" />
<MemberSignature Language="ILAsm" Value=".method public static hidebysig class System.Collections.Immutable.ImmutableDictionary`2<string, class Microsoft.ML.Data.MulticlassClassificationMetricsStatistics> PermutationFeatureImportance(class Microsoft.ML.MulticlassClassificationCatalog catalog, class Microsoft.ML.ITransformer model, class Microsoft.ML.IDataView data, string labelColumnName, bool useFeatureWeightFilter, valuetype System.Nullable`1<int32> numberOfExamplesToUse, int32 permutationCount) cil managed" />
<MemberSignature Language="DocId" Value="M:Microsoft.ML.PermutationFeatureImportanceExtensions.PermutationFeatureImportance(Microsoft.ML.MulticlassClassificationCatalog,Microsoft.ML.ITransformer,Microsoft.ML.IDataView,System.String,System.Boolean,System.Nullable{System.Int32},System.Int32)" />
<MemberSignature Language="VB.NET" Value="<Extension()>
Public Function PermutationFeatureImportance (catalog As MulticlassClassificationCatalog, model As ITransformer, data As IDataView, Optional labelColumnName As String = "Label", Optional useFeatureWeightFilter As Boolean = false, Optional numberOfExamplesToUse As Nullable(Of Integer) = Nothing, Optional permutationCount As Integer = 1) As ImmutableDictionary(Of String, MulticlassClassificationMetricsStatistics)" />
<MemberSignature Language="F#" Value="static member PermutationFeatureImportance : Microsoft.ML.MulticlassClassificationCatalog * Microsoft.ML.ITransformer * Microsoft.ML.IDataView * string * bool * Nullable<int> * int -> System.Collections.Immutable.ImmutableDictionary<string, Microsoft.ML.Data.MulticlassClassificationMetricsStatistics>" Usage="Microsoft.ML.PermutationFeatureImportanceExtensions.PermutationFeatureImportance (catalog, model, data, labelColumnName, useFeatureWeightFilter, numberOfExamplesToUse, permutationCount)" />
<MemberType>Method</MemberType>
<AssemblyInfo>
<AssemblyName>Microsoft.ML.Transforms</AssemblyName>
<AssemblyVersion>1.0.0.0</AssemblyVersion>
</AssemblyInfo>
<ReturnValue>
<ReturnType>System.Collections.Immutable.ImmutableDictionary<System.String,Microsoft.ML.Data.MulticlassClassificationMetricsStatistics></ReturnType>
</ReturnValue>
<Parameters>
<Parameter Name="catalog" Type="Microsoft.ML.MulticlassClassificationCatalog" RefType="this" Index="0" FrameworkAlternate="ml-dotnet;ml-dotnet-1.7.0;ml-dotnet-2.0.0" />
<Parameter Name="model" Type="Microsoft.ML.ITransformer" Index="1" FrameworkAlternate="ml-dotnet;ml-dotnet-1.7.0;ml-dotnet-2.0.0" />
<Parameter Name="data" Type="Microsoft.ML.IDataView" Index="2" FrameworkAlternate="ml-dotnet;ml-dotnet-1.7.0;ml-dotnet-2.0.0" />
<Parameter Name="labelColumnName" Type="System.String" Index="3" FrameworkAlternate="ml-dotnet;ml-dotnet-1.7.0;ml-dotnet-2.0.0" />
<Parameter Name="useFeatureWeightFilter" Type="System.Boolean" Index="4" FrameworkAlternate="ml-dotnet;ml-dotnet-1.7.0;ml-dotnet-2.0.0" />
<Parameter Name="numberOfExamplesToUse" Type="System.Nullable<System.Int32>" Index="5" FrameworkAlternate="ml-dotnet;ml-dotnet-1.7.0;ml-dotnet-2.0.0" />
<Parameter Name="permutationCount" Type="System.Int32" Index="6" FrameworkAlternate="ml-dotnet;ml-dotnet-1.7.0;ml-dotnet-2.0.0" />
</Parameters>
<Docs>
<param name="catalog">The multiclass classification catalog.</param>
<param name="model">The model on which to evaluate feature importance.</param>
<param name="data">The evaluation data set.</param>
<param name="labelColumnName">Label column name. The column data must be <see cref="T:Microsoft.ML.Data.KeyDataViewType" />.</param>
<param name="useFeatureWeightFilter">Use features weight to pre-filter features.</param>
<param name="numberOfExamplesToUse">Limit the number of examples to evaluate on. <cref langword="null" /> means up to ~2 bln examples from <paramref param="data" /> will be used.</param>
<param name="permutationCount">The number of permutations to perform.</param>
<summary>
Permutation Feature Importance (PFI) for MulticlassClassification.
</summary>
<returns>Dictionary mapping each feature to its per-feature 'contributions' to the score.</returns>
<remarks>
<para>
Permutation feature importance (PFI) is a technique to determine the global importance of features in a trained
machine learning model. PFI is a simple yet powerful technique motivated by Breiman in his Random Forest paper, section 10
(Breiman. <a href="https://www.stat.berkeley.edu/~breiman/randomforest2001.pdf">"Random Forests."</a> Machine Learning, 2001.)
The advantage of the PFI method is that it is model agnostic -- it works with any model that can be
evaluated -- and it can use any dataset, not just the training set, to compute feature importance metrics.
</para>
<para>
PFI works by taking a labeled dataset, choosing a feature, and permuting the values
for that feature across all the examples, so that each example now has a random value for the feature and
the original values for all other features. The evaluation metric (e.g. micro-accuracy) is then calculated
for this modified dataset, and the change in the evaluation metric from the original dataset is computed.
The larger the change in the evaluation metric, the more important the feature is to the model.
PFI works by performing this permutation analysis across all the features of a model, one after another.
</para>
<para>
In this implementation, PFI computes the change in all possible multiclass classification evaluation metrics for each feature, and an
<see cref="T:System.Collections.Immutable.ImmutableArray" /> of <see cref="T:Microsoft.ML.Data.MulticlassClassificationMetrics" /> objects is returned. See the sample below for an
example of working with these results to analyze the feature importance of a model.
</para>
</remarks>
<example>
<format type="text/markdown"><![CDATA[
[!code-csharp[PermutationFeatureImportance](~/../docs/samples/docs/samples/Microsoft.ML.Samples/Dynamic/Trainers/MulticlassClassification/PermutationFeatureImportance.cs)]
]]></format>
</example>
</Docs>
</Member>
<Member MemberName="PermutationFeatureImportance">
<MemberSignature Language="C#" Value="public static System.Collections.Immutable.ImmutableDictionary<string,Microsoft.ML.Data.RegressionMetricsStatistics> PermutationFeatureImportance (this Microsoft.ML.RegressionCatalog catalog, Microsoft.ML.ITransformer model, Microsoft.ML.IDataView data, string labelColumnName = "Label", bool useFeatureWeightFilter = false, int? numberOfExamplesToUse = default, int permutationCount = 1);" />
<MemberSignature Language="ILAsm" Value=".method public static hidebysig class System.Collections.Immutable.ImmutableDictionary`2<string, class Microsoft.ML.Data.RegressionMetricsStatistics> PermutationFeatureImportance(class Microsoft.ML.RegressionCatalog catalog, class Microsoft.ML.ITransformer model, class Microsoft.ML.IDataView data, string labelColumnName, bool useFeatureWeightFilter, valuetype System.Nullable`1<int32> numberOfExamplesToUse, int32 permutationCount) cil managed" />
<MemberSignature Language="DocId" Value="M:Microsoft.ML.PermutationFeatureImportanceExtensions.PermutationFeatureImportance(Microsoft.ML.RegressionCatalog,Microsoft.ML.ITransformer,Microsoft.ML.IDataView,System.String,System.Boolean,System.Nullable{System.Int32},System.Int32)" />
<MemberSignature Language="VB.NET" Value="<Extension()>
Public Function PermutationFeatureImportance (catalog As RegressionCatalog, model As ITransformer, data As IDataView, Optional labelColumnName As String = "Label", Optional useFeatureWeightFilter As Boolean = false, Optional numberOfExamplesToUse As Nullable(Of Integer) = Nothing, Optional permutationCount As Integer = 1) As ImmutableDictionary(Of String, RegressionMetricsStatistics)" />
<MemberSignature Language="F#" Value="static member PermutationFeatureImportance : Microsoft.ML.RegressionCatalog * Microsoft.ML.ITransformer * Microsoft.ML.IDataView * string * bool * Nullable<int> * int -> System.Collections.Immutable.ImmutableDictionary<string, Microsoft.ML.Data.RegressionMetricsStatistics>" Usage="Microsoft.ML.PermutationFeatureImportanceExtensions.PermutationFeatureImportance (catalog, model, data, labelColumnName, useFeatureWeightFilter, numberOfExamplesToUse, permutationCount)" />
<MemberType>Method</MemberType>
<AssemblyInfo>
<AssemblyName>Microsoft.ML.Transforms</AssemblyName>
<AssemblyVersion>1.0.0.0</AssemblyVersion>
</AssemblyInfo>
<ReturnValue>
<ReturnType>System.Collections.Immutable.ImmutableDictionary<System.String,Microsoft.ML.Data.RegressionMetricsStatistics></ReturnType>
</ReturnValue>
<Parameters>
<Parameter Name="catalog" Type="Microsoft.ML.RegressionCatalog" RefType="this" Index="0" FrameworkAlternate="ml-dotnet;ml-dotnet-1.7.0;ml-dotnet-2.0.0" />
<Parameter Name="model" Type="Microsoft.ML.ITransformer" Index="1" FrameworkAlternate="ml-dotnet;ml-dotnet-1.7.0;ml-dotnet-2.0.0" />
<Parameter Name="data" Type="Microsoft.ML.IDataView" Index="2" FrameworkAlternate="ml-dotnet;ml-dotnet-1.7.0;ml-dotnet-2.0.0" />
<Parameter Name="labelColumnName" Type="System.String" Index="3" FrameworkAlternate="ml-dotnet;ml-dotnet-1.7.0;ml-dotnet-2.0.0" />
<Parameter Name="useFeatureWeightFilter" Type="System.Boolean" Index="4" FrameworkAlternate="ml-dotnet;ml-dotnet-1.7.0;ml-dotnet-2.0.0" />
<Parameter Name="numberOfExamplesToUse" Type="System.Nullable<System.Int32>" Index="5" FrameworkAlternate="ml-dotnet;ml-dotnet-1.7.0;ml-dotnet-2.0.0" />
<Parameter Name="permutationCount" Type="System.Int32" Index="6" FrameworkAlternate="ml-dotnet;ml-dotnet-1.7.0;ml-dotnet-2.0.0" />
</Parameters>
<Docs>
<param name="catalog">The regression catalog.</param>
<param name="model">The model on which to evaluate feature importance.</param>
<param name="data">The evaluation data set.</param>
<param name="labelColumnName">Label column name. The column data must be <see cref="T:System.Single" />.</param>
<param name="useFeatureWeightFilter">Use features weight to pre-filter features.</param>
<param name="numberOfExamplesToUse">Limit the number of examples to evaluate on. <cref langword="null" /> means up to ~2 bln examples from <paramref param="data" /> will be used.</param>
<param name="permutationCount">The number of permutations to perform.</param>
<summary>
Permutation Feature Importance (PFI) for Regression.
</summary>
<returns>Dictionary mapping each feature to its per-feature 'contributions' to the score.</returns>
<remarks>
<para>
Permutation feature importance (PFI) is a technique to determine the global importance of features in a trained
machine learning model. PFI is a simple yet powerful technique motivated by Breiman in his Random Forest paper, section 10
(Breiman. <a href="https://www.stat.berkeley.edu/~breiman/randomforest2001.pdf">"Random Forests."</a> Machine Learning, 2001.)
The advantage of the PFI method is that it is model agnostic -- it works with any model that can be
evaluated -- and it can use any dataset, not just the training set, to compute feature importance metrics.
</para>
<para>
PFI works by taking a labeled dataset, choosing a feature, and permuting the values
for that feature across all the examples, so that each example now has a random value for the feature and
the original values for all other features. The evaluation metric (e.g. R-squared) is then calculated
for this modified dataset, and the change in the evaluation metric from the original dataset is computed.
The larger the change in the evaluation metric, the more important the feature is to the model.
PFI works by performing this permutation analysis across all the features of a model, one after another.
</para>
<para>
In this implementation, PFI computes the change in all possible regression evaluation metrics for each feature, and an
<see cref="T:System.Collections.Immutable.ImmutableArray" /> of <see cref="T:Microsoft.ML.Data.RegressionMetrics" /> objects is returned. See the sample below for an
example of working with these results to analyze the feature importance of a model.
</para>
</remarks>
<example>
<format type="text/markdown"><![CDATA[
[!code-csharp[PermutationFeatureImportance](~/../docs/samples/docs/samples/Microsoft.ML.Samples/Dynamic/Trainers/Regression/PermutationFeatureImportance.cs)]
]]></format>
</example>
</Docs>
</Member>
<Member MemberName="PermutationFeatureImportance">
<MemberSignature Language="C#" Value="public static System.Collections.Immutable.ImmutableDictionary<string,Microsoft.ML.Data.RankingMetricsStatistics> PermutationFeatureImportance (this Microsoft.ML.RankingCatalog catalog, Microsoft.ML.ITransformer model, Microsoft.ML.IDataView data, string labelColumnName = "Label", string rowGroupColumnName = "GroupId", bool useFeatureWeightFilter = false, int? numberOfExamplesToUse = default, int permutationCount = 1);" />
<MemberSignature Language="ILAsm" Value=".method public static hidebysig class System.Collections.Immutable.ImmutableDictionary`2<string, class Microsoft.ML.Data.RankingMetricsStatistics> PermutationFeatureImportance(class Microsoft.ML.RankingCatalog catalog, class Microsoft.ML.ITransformer model, class Microsoft.ML.IDataView data, string labelColumnName, string rowGroupColumnName, bool useFeatureWeightFilter, valuetype System.Nullable`1<int32> numberOfExamplesToUse, int32 permutationCount) cil managed" />
<MemberSignature Language="DocId" Value="M:Microsoft.ML.PermutationFeatureImportanceExtensions.PermutationFeatureImportance(Microsoft.ML.RankingCatalog,Microsoft.ML.ITransformer,Microsoft.ML.IDataView,System.String,System.String,System.Boolean,System.Nullable{System.Int32},System.Int32)" />
<MemberSignature Language="VB.NET" Value="<Extension()>
Public Function PermutationFeatureImportance (catalog As RankingCatalog, model As ITransformer, data As IDataView, Optional labelColumnName As String = "Label", Optional rowGroupColumnName As String = "GroupId", Optional useFeatureWeightFilter As Boolean = false, Optional numberOfExamplesToUse As Nullable(Of Integer) = Nothing, Optional permutationCount As Integer = 1) As ImmutableDictionary(Of String, RankingMetricsStatistics)" />
<MemberSignature Language="F#" Value="static member PermutationFeatureImportance : Microsoft.ML.RankingCatalog * Microsoft.ML.ITransformer * Microsoft.ML.IDataView * string * string * bool * Nullable<int> * int -> System.Collections.Immutable.ImmutableDictionary<string, Microsoft.ML.Data.RankingMetricsStatistics>" Usage="Microsoft.ML.PermutationFeatureImportanceExtensions.PermutationFeatureImportance (catalog, model, data, labelColumnName, rowGroupColumnName, useFeatureWeightFilter, numberOfExamplesToUse, permutationCount)" />
<MemberType>Method</MemberType>
<AssemblyInfo>
<AssemblyName>Microsoft.ML.Transforms</AssemblyName>
<AssemblyVersion>1.0.0.0</AssemblyVersion>
</AssemblyInfo>
<ReturnValue>
<ReturnType>System.Collections.Immutable.ImmutableDictionary<System.String,Microsoft.ML.Data.RankingMetricsStatistics></ReturnType>
</ReturnValue>
<Parameters>
<Parameter Name="catalog" Type="Microsoft.ML.RankingCatalog" RefType="this" Index="0" FrameworkAlternate="ml-dotnet;ml-dotnet-1.7.0;ml-dotnet-2.0.0" />
<Parameter Name="model" Type="Microsoft.ML.ITransformer" Index="1" FrameworkAlternate="ml-dotnet;ml-dotnet-1.7.0;ml-dotnet-2.0.0" />
<Parameter Name="data" Type="Microsoft.ML.IDataView" Index="2" FrameworkAlternate="ml-dotnet;ml-dotnet-1.7.0;ml-dotnet-2.0.0" />
<Parameter Name="labelColumnName" Type="System.String" Index="3" FrameworkAlternate="ml-dotnet;ml-dotnet-1.7.0;ml-dotnet-2.0.0" />
<Parameter Name="rowGroupColumnName" Type="System.String" Index="4" FrameworkAlternate="ml-dotnet;ml-dotnet-1.7.0;ml-dotnet-2.0.0" />
<Parameter Name="useFeatureWeightFilter" Type="System.Boolean" Index="5" FrameworkAlternate="ml-dotnet;ml-dotnet-1.7.0;ml-dotnet-2.0.0" />
<Parameter Name="numberOfExamplesToUse" Type="System.Nullable<System.Int32>" Index="6" FrameworkAlternate="ml-dotnet;ml-dotnet-1.7.0;ml-dotnet-2.0.0" />
<Parameter Name="permutationCount" Type="System.Int32" Index="7" FrameworkAlternate="ml-dotnet;ml-dotnet-1.7.0;ml-dotnet-2.0.0" />
</Parameters>
<Docs>
<param name="catalog">The ranking catalog.</param>
<param name="model">The model on which to evaluate feature importance.</param>
<param name="data">The evaluation data set.</param>
<param name="labelColumnName">Label column name. The column data must be <see cref="T:System.Single" /> or <see cref="T:Microsoft.ML.Data.KeyDataViewType" />.</param>
<param name="rowGroupColumnName">GroupId column name</param>
<param name="useFeatureWeightFilter">Use features weight to pre-filter features.</param>
<param name="numberOfExamplesToUse">Limit the number of examples to evaluate on. <cref langword="null" /> means up to ~2 bln examples from <paramref param="data" /> will be used.</param>
<param name="permutationCount">The number of permutations to perform.</param>
<summary>
Permutation Feature Importance (PFI) for Ranking.
</summary>
<returns>Dictionary mapping each feature to its per-feature 'contributions' to the score.</returns>
<remarks>
<para>
Permutation feature importance (PFI) is a technique to determine the global importance of features in a trained
machine learning model. PFI is a simple yet powerful technique motivated by Breiman in his Random Forest paper, section 10
(Breiman. <a href="https://www.stat.berkeley.edu/~breiman/randomforest2001.pdf">"Random Forests."</a> Machine Learning, 2001.)
The advantage of the PFI method is that it is model agnostic -- it works with any model that can be
evaluated -- and it can use any dataset, not just the training set, to compute feature importance metrics.
</para>
<para>
PFI works by taking a labeled dataset, choosing a feature, and permuting the values
for that feature across all the examples, so that each example now has a random value for the feature and
the original values for all other features. The evaluation metric (e.g. NDCG) is then calculated
for this modified dataset, and the change in the evaluation metric from the original dataset is computed.
The larger the change in the evaluation metric, the more important the feature is to the model.
PFI works by performing this permutation analysis across all the features of a model, one after another.
</para>
<para>
In this implementation, PFI computes the change in all possible ranking evaluation metrics for each feature, and an
<see cref="T:System.Collections.Immutable.ImmutableArray" /> of <see cref="T:Microsoft.ML.Data.RankingMetrics" /> objects is returned. See the sample below for an
example of working with these results to analyze the feature importance of a model.
</para>
</remarks>
<example>
<format type="text/markdown"><![CDATA[
[!code-csharp[PermutationFeatureImportance](~/../docs/samples/docs/samples/Microsoft.ML.Samples/Dynamic/Trainers/Ranking/PermutationFeatureImportance.cs)]
]]></format>
</example>
</Docs>
</Member>
<Member MemberName="PermutationFeatureImportance<TModel>">
<MemberSignature Language="C#" Value="public static System.Collections.Immutable.ImmutableArray<Microsoft.ML.Data.BinaryClassificationMetricsStatistics> PermutationFeatureImportance<TModel> (this Microsoft.ML.BinaryClassificationCatalog catalog, Microsoft.ML.ISingleFeaturePredictionTransformer<TModel> predictionTransformer, Microsoft.ML.IDataView data, string labelColumnName = "Label", bool useFeatureWeightFilter = false, int? numberOfExamplesToUse = default, int permutationCount = 1) where TModel : class;" />
<MemberSignature Language="ILAsm" Value=".method public static hidebysig valuetype System.Collections.Immutable.ImmutableArray`1<class Microsoft.ML.Data.BinaryClassificationMetricsStatistics> PermutationFeatureImportance<class TModel>(class Microsoft.ML.BinaryClassificationCatalog catalog, class Microsoft.ML.ISingleFeaturePredictionTransformer`1<!!TModel> predictionTransformer, class Microsoft.ML.IDataView data, string labelColumnName, bool useFeatureWeightFilter, valuetype System.Nullable`1<int32> numberOfExamplesToUse, int32 permutationCount) cil managed" />
<MemberSignature Language="DocId" Value="M:Microsoft.ML.PermutationFeatureImportanceExtensions.PermutationFeatureImportance``1(Microsoft.ML.BinaryClassificationCatalog,Microsoft.ML.ISingleFeaturePredictionTransformer{``0},Microsoft.ML.IDataView,System.String,System.Boolean,System.Nullable{System.Int32},System.Int32)" />
<MemberSignature Language="VB.NET" Value="<Extension()>
Public Function PermutationFeatureImportance(Of TModel As Class) (catalog As BinaryClassificationCatalog, predictionTransformer As ISingleFeaturePredictionTransformer(Of TModel), data As IDataView, Optional labelColumnName As String = "Label", Optional useFeatureWeightFilter As Boolean = false, Optional numberOfExamplesToUse As Nullable(Of Integer) = Nothing, Optional permutationCount As Integer = 1) As ImmutableArray(Of BinaryClassificationMetricsStatistics)" />
<MemberSignature Language="F#" Value="static member PermutationFeatureImportance : Microsoft.ML.BinaryClassificationCatalog * Microsoft.ML.ISingleFeaturePredictionTransformer<'Model (requires 'Model : null)> * Microsoft.ML.IDataView * string * bool * Nullable<int> * int -> System.Collections.Immutable.ImmutableArray<Microsoft.ML.Data.BinaryClassificationMetricsStatistics> (requires 'Model : null)" Usage="Microsoft.ML.PermutationFeatureImportanceExtensions.PermutationFeatureImportance (catalog, predictionTransformer, data, labelColumnName, useFeatureWeightFilter, numberOfExamplesToUse, permutationCount)" />
<MemberType>Method</MemberType>
<AssemblyInfo>
<AssemblyName>Microsoft.ML.Transforms</AssemblyName>
<AssemblyVersion>1.0.0.0</AssemblyVersion>
</AssemblyInfo>
<ReturnValue>
<ReturnType>System.Collections.Immutable.ImmutableArray<Microsoft.ML.Data.BinaryClassificationMetricsStatistics></ReturnType>
</ReturnValue>
<TypeParameters>
<TypeParameter Name="TModel">
<Constraints>
<ParameterAttribute>ReferenceTypeConstraint</ParameterAttribute>
</Constraints>
</TypeParameter>
</TypeParameters>
<Parameters>
<Parameter Name="catalog" Type="Microsoft.ML.BinaryClassificationCatalog" RefType="this" />
<Parameter Name="predictionTransformer" Type="Microsoft.ML.ISingleFeaturePredictionTransformer<TModel>" />
<Parameter Name="data" Type="Microsoft.ML.IDataView" />
<Parameter Name="labelColumnName" Type="System.String" />
<Parameter Name="useFeatureWeightFilter" Type="System.Boolean" />
<Parameter Name="numberOfExamplesToUse" Type="System.Nullable<System.Int32>" />
<Parameter Name="permutationCount" Type="System.Int32" />
</Parameters>
<Docs>
<typeparam name="TModel">To be added.</typeparam>
<param name="catalog">The binary classification catalog.</param>
<param name="predictionTransformer">The model on which to evaluate feature importance.</param>
<param name="data">The evaluation data set.</param>
<param name="labelColumnName">Label column name. The column data must be <see cref="T:System.Boolean" />.</param>
<param name="useFeatureWeightFilter">Use features weight to pre-filter features.</param>
<param name="numberOfExamplesToUse">Limit the number of examples to evaluate on. <cref langword="null" /> means up to ~2 bln examples from <paramref param="data" /> will be used.</param>
<param name="permutationCount">The number of permutations to perform.</param>
<summary>
Permutation Feature Importance (PFI) for Binary Classification.
</summary>
<returns>Array of per-feature 'contributions' to the score.</returns>
<remarks>
<para>
Permutation feature importance (PFI) is a technique to determine the global importance of features in a trained
machine learning model. PFI is a simple yet powerful technique motivated by Breiman in his Random Forest paper, section 10
(Breiman. <a href="https://www.stat.berkeley.edu/~breiman/randomforest2001.pdf">"Random Forests."</a> Machine Learning, 2001.)
The advantage of the PFI method is that it is model agnostic -- it works with any model that can be
evaluated -- and it can use any dataset, not just the training set, to compute feature importance metrics.
</para>
<para>
PFI works by taking a labeled dataset, choosing a feature, and permuting the values
for that feature across all the examples, so that each example now has a random value for the feature and
the original values for all other features. The evaluation metric (e.g. AUC) is then calculated
for this modified dataset, and the change in the evaluation metric from the original dataset is computed.
The larger the change in the evaluation metric, the more important the feature is to the model.
PFI works by performing this permutation analysis across all the features of a model, one after another.
</para>
<para>
In this implementation, PFI computes the change in all possible binary classification evaluation metrics for each feature, and an
<see cref="T:System.Collections.Immutable.ImmutableArray" /> of <see cref="T:Microsoft.ML.Data.BinaryClassificationMetrics" /> objects is returned. See the sample below for an
example of working with these results to analyze the feature importance of a model.
</para>
</remarks>
<example>
<format type="text/markdown"><![CDATA[
[!code-csharp[PermutationFeatureImportance](~/../docs/samples/docs/samples/Microsoft.ML.Samples/Dynamic/Trainers/BinaryClassification/PermutationFeatureImportance.cs)]
]]></format>
</example>
</Docs>
</Member>
<Member MemberName="PermutationFeatureImportance<TModel>">
<MemberSignature Language="C#" Value="public static System.Collections.Immutable.ImmutableArray<Microsoft.ML.Data.MulticlassClassificationMetricsStatistics> PermutationFeatureImportance<TModel> (this Microsoft.ML.MulticlassClassificationCatalog catalog, Microsoft.ML.ISingleFeaturePredictionTransformer<TModel> predictionTransformer, Microsoft.ML.IDataView data, string labelColumnName = "Label", bool useFeatureWeightFilter = false, int? numberOfExamplesToUse = default, int permutationCount = 1) where TModel : class;" />
<MemberSignature Language="ILAsm" Value=".method public static hidebysig valuetype System.Collections.Immutable.ImmutableArray`1<class Microsoft.ML.Data.MulticlassClassificationMetricsStatistics> PermutationFeatureImportance<class TModel>(class Microsoft.ML.MulticlassClassificationCatalog catalog, class Microsoft.ML.ISingleFeaturePredictionTransformer`1<!!TModel> predictionTransformer, class Microsoft.ML.IDataView data, string labelColumnName, bool useFeatureWeightFilter, valuetype System.Nullable`1<int32> numberOfExamplesToUse, int32 permutationCount) cil managed" />
<MemberSignature Language="DocId" Value="M:Microsoft.ML.PermutationFeatureImportanceExtensions.PermutationFeatureImportance``1(Microsoft.ML.MulticlassClassificationCatalog,Microsoft.ML.ISingleFeaturePredictionTransformer{``0},Microsoft.ML.IDataView,System.String,System.Boolean,System.Nullable{System.Int32},System.Int32)" />
<MemberSignature Language="VB.NET" Value="<Extension()>
Public Function PermutationFeatureImportance(Of TModel As Class) (catalog As MulticlassClassificationCatalog, predictionTransformer As ISingleFeaturePredictionTransformer(Of TModel), data As IDataView, Optional labelColumnName As String = "Label", Optional useFeatureWeightFilter As Boolean = false, Optional numberOfExamplesToUse As Nullable(Of Integer) = Nothing, Optional permutationCount As Integer = 1) As ImmutableArray(Of MulticlassClassificationMetricsStatistics)" />
<MemberSignature Language="F#" Value="static member PermutationFeatureImportance : Microsoft.ML.MulticlassClassificationCatalog * Microsoft.ML.ISingleFeaturePredictionTransformer<'Model (requires 'Model : null)> * Microsoft.ML.IDataView * string * bool * Nullable<int> * int -> System.Collections.Immutable.ImmutableArray<Microsoft.ML.Data.MulticlassClassificationMetricsStatistics> (requires 'Model : null)" Usage="Microsoft.ML.PermutationFeatureImportanceExtensions.PermutationFeatureImportance (catalog, predictionTransformer, data, labelColumnName, useFeatureWeightFilter, numberOfExamplesToUse, permutationCount)" />
<MemberType>Method</MemberType>
<AssemblyInfo>
<AssemblyName>Microsoft.ML.Transforms</AssemblyName>
<AssemblyVersion>1.0.0.0</AssemblyVersion>
</AssemblyInfo>
<ReturnValue>
<ReturnType>System.Collections.Immutable.ImmutableArray<Microsoft.ML.Data.MulticlassClassificationMetricsStatistics></ReturnType>
</ReturnValue>
<TypeParameters>
<TypeParameter Name="TModel">
<Constraints>
<ParameterAttribute>ReferenceTypeConstraint</ParameterAttribute>
</Constraints>
</TypeParameter>
</TypeParameters>
<Parameters>
<Parameter Name="catalog" Type="Microsoft.ML.MulticlassClassificationCatalog" RefType="this" />
<Parameter Name="predictionTransformer" Type="Microsoft.ML.ISingleFeaturePredictionTransformer<TModel>" />
<Parameter Name="data" Type="Microsoft.ML.IDataView" />
<Parameter Name="labelColumnName" Type="System.String" />
<Parameter Name="useFeatureWeightFilter" Type="System.Boolean" />
<Parameter Name="numberOfExamplesToUse" Type="System.Nullable<System.Int32>" />
<Parameter Name="permutationCount" Type="System.Int32" />
</Parameters>
<Docs>
<typeparam name="TModel">To be added.</typeparam>
<param name="catalog">The multiclass classification catalog.</param>
<param name="predictionTransformer">The model on which to evaluate feature importance.</param>
<param name="data">The evaluation data set.</param>
<param name="labelColumnName">Label column name. The column data must be <see cref="T:Microsoft.ML.Data.KeyDataViewType" />.</param>
<param name="useFeatureWeightFilter">Use features weight to pre-filter features.</param>
<param name="numberOfExamplesToUse">Limit the number of examples to evaluate on. <cref langword="null" /> means up to ~2 bln examples from <paramref param="data" /> will be used.</param>
<param name="permutationCount">The number of permutations to perform.</param>
<summary>
Permutation Feature Importance (PFI) for MulticlassClassification.
</summary>
<returns>Array of per-feature 'contributions' to the score.</returns>
<remarks>
<para>
Permutation feature importance (PFI) is a technique to determine the global importance of features in a trained
machine learning model. PFI is a simple yet powerful technique motivated by Breiman in his Random Forest paper, section 10
(Breiman. <a href="https://www.stat.berkeley.edu/~breiman/randomforest2001.pdf">"Random Forests."</a> Machine Learning, 2001.)
The advantage of the PFI method is that it is model agnostic -- it works with any model that can be
evaluated -- and it can use any dataset, not just the training set, to compute feature importance metrics.
</para>
<para>
PFI works by taking a labeled dataset, choosing a feature, and permuting the values
for that feature across all the examples, so that each example now has a random value for the feature and
the original values for all other features. The evaluation metric (e.g. micro-accuracy) is then calculated
for this modified dataset, and the change in the evaluation metric from the original dataset is computed.
The larger the change in the evaluation metric, the more important the feature is to the model.
PFI works by performing this permutation analysis across all the features of a model, one after another.
</para>
<para>
In this implementation, PFI computes the change in all possible multiclass classification evaluation metrics for each feature, and an
<see cref="T:System.Collections.Immutable.ImmutableArray" /> of <see cref="T:Microsoft.ML.Data.MulticlassClassificationMetrics" /> objects is returned. See the sample below for an
example of working with these results to analyze the feature importance of a model.
</para>
</remarks>
<example>
<format type="text/markdown"><![CDATA[
[!code-csharp[PermutationFeatureImportance](~/../docs/samples/docs/samples/Microsoft.ML.Samples/Dynamic/Trainers/MulticlassClassification/PermutationFeatureImportance.cs)]
]]></format>
</example>
</Docs>
</Member>
<Member MemberName="PermutationFeatureImportance<TModel>">
<MemberSignature Language="C#" Value="public static System.Collections.Immutable.ImmutableArray<Microsoft.ML.Data.RegressionMetricsStatistics> PermutationFeatureImportance<TModel> (this Microsoft.ML.RegressionCatalog catalog, Microsoft.ML.ISingleFeaturePredictionTransformer<TModel> predictionTransformer, Microsoft.ML.IDataView data, string labelColumnName = "Label", bool useFeatureWeightFilter = false, int? numberOfExamplesToUse = default, int permutationCount = 1) where TModel : class;" />
<MemberSignature Language="ILAsm" Value=".method public static hidebysig valuetype System.Collections.Immutable.ImmutableArray`1<class Microsoft.ML.Data.RegressionMetricsStatistics> PermutationFeatureImportance<class TModel>(class Microsoft.ML.RegressionCatalog catalog, class Microsoft.ML.ISingleFeaturePredictionTransformer`1<!!TModel> predictionTransformer, class Microsoft.ML.IDataView data, string labelColumnName, bool useFeatureWeightFilter, valuetype System.Nullable`1<int32> numberOfExamplesToUse, int32 permutationCount) cil managed" />
<MemberSignature Language="DocId" Value="M:Microsoft.ML.PermutationFeatureImportanceExtensions.PermutationFeatureImportance``1(Microsoft.ML.RegressionCatalog,Microsoft.ML.ISingleFeaturePredictionTransformer{``0},Microsoft.ML.IDataView,System.String,System.Boolean,System.Nullable{System.Int32},System.Int32)" />
<MemberSignature Language="VB.NET" Value="<Extension()>
Public Function PermutationFeatureImportance(Of TModel As Class) (catalog As RegressionCatalog, predictionTransformer As ISingleFeaturePredictionTransformer(Of TModel), data As IDataView, Optional labelColumnName As String = "Label", Optional useFeatureWeightFilter As Boolean = false, Optional numberOfExamplesToUse As Nullable(Of Integer) = Nothing, Optional permutationCount As Integer = 1) As ImmutableArray(Of RegressionMetricsStatistics)" />
<MemberSignature Language="F#" Value="static member PermutationFeatureImportance : Microsoft.ML.RegressionCatalog * Microsoft.ML.ISingleFeaturePredictionTransformer<'Model (requires 'Model : null)> * Microsoft.ML.IDataView * string * bool * Nullable<int> * int -> System.Collections.Immutable.ImmutableArray<Microsoft.ML.Data.RegressionMetricsStatistics> (requires 'Model : null)" Usage="Microsoft.ML.PermutationFeatureImportanceExtensions.PermutationFeatureImportance (catalog, predictionTransformer, data, labelColumnName, useFeatureWeightFilter, numberOfExamplesToUse, permutationCount)" />
<MemberType>Method</MemberType>
<AssemblyInfo>
<AssemblyName>Microsoft.ML.Transforms</AssemblyName>
<AssemblyVersion>1.0.0.0</AssemblyVersion>
</AssemblyInfo>
<ReturnValue>
<ReturnType>System.Collections.Immutable.ImmutableArray<Microsoft.ML.Data.RegressionMetricsStatistics></ReturnType>
</ReturnValue>
<TypeParameters>
<TypeParameter Name="TModel">
<Constraints>
<ParameterAttribute>ReferenceTypeConstraint</ParameterAttribute>
</Constraints>
</TypeParameter>
</TypeParameters>
<Parameters>
<Parameter Name="catalog" Type="Microsoft.ML.RegressionCatalog" RefType="this" />
<Parameter Name="predictionTransformer" Type="Microsoft.ML.ISingleFeaturePredictionTransformer<TModel>" />
<Parameter Name="data" Type="Microsoft.ML.IDataView" />
<Parameter Name="labelColumnName" Type="System.String" />
<Parameter Name="useFeatureWeightFilter" Type="System.Boolean" />
<Parameter Name="numberOfExamplesToUse" Type="System.Nullable<System.Int32>" />
<Parameter Name="permutationCount" Type="System.Int32" />
</Parameters>
<Docs>
<typeparam name="TModel">To be added.</typeparam>
<param name="catalog">The regression catalog.</param>
<param name="predictionTransformer">The model on which to evaluate feature importance.</param>
<param name="data">The evaluation data set.</param>
<param name="labelColumnName">Label column name. The column data must be <see cref="T:System.Single" />.</param>
<param name="useFeatureWeightFilter">Use features weight to pre-filter features.</param>
<param name="numberOfExamplesToUse">Limit the number of examples to evaluate on. <cref langword="null" /> means up to ~2 bln examples from <paramref param="data" /> will be used.</param>
<param name="permutationCount">The number of permutations to perform.</param>
<summary>
Permutation Feature Importance (PFI) for Regression.
</summary>
<returns>Array of per-feature 'contributions' to the score.</returns>
<remarks>
<para>
Permutation feature importance (PFI) is a technique to determine the global importance of features in a trained
machine learning model. PFI is a simple yet powerful technique motivated by Breiman in his Random Forest paper, section 10
(Breiman. <a href="https://www.stat.berkeley.edu/~breiman/randomforest2001.pdf">"Random Forests."</a> Machine Learning, 2001.)
The advantage of the PFI method is that it is model agnostic -- it works with any model that can be
evaluated -- and it can use any dataset, not just the training set, to compute feature importance metrics.
</para>
<para>
PFI works by taking a labeled dataset, choosing a feature, and permuting the values
for that feature across all the examples, so that each example now has a random value for the feature and
the original values for all other features. The evaluation metric (e.g. R-squared) is then calculated
for this modified dataset, and the change in the evaluation metric from the original dataset is computed.
The larger the change in the evaluation metric, the more important the feature is to the model.
PFI works by performing this permutation analysis across all the features of a model, one after another.
</para>
<para>
In this implementation, PFI computes the change in all possible regression evaluation metrics for each feature, and an
<see cref="T:System.Collections.Immutable.ImmutableArray" /> of <see cref="T:Microsoft.ML.Data.RegressionMetrics" /> objects is returned. See the sample below for an
example of working with these results to analyze the feature importance of a model.
</para>
</remarks>
<example>
<format type="text/markdown"><![CDATA[
[!code-csharp[PermutationFeatureImportance](~/../docs/samples/docs/samples/Microsoft.ML.Samples/Dynamic/Trainers/Regression/PermutationFeatureImportance.cs)]
]]></format>
</example>
</Docs>
</Member>
<Member MemberName="PermutationFeatureImportance<TModel>">
<MemberSignature Language="C#" Value="public static System.Collections.Immutable.ImmutableArray<Microsoft.ML.Data.RankingMetricsStatistics> PermutationFeatureImportance<TModel> (this Microsoft.ML.RankingCatalog catalog, Microsoft.ML.ISingleFeaturePredictionTransformer<TModel> predictionTransformer, Microsoft.ML.IDataView data, string labelColumnName = "Label", string rowGroupColumnName = "GroupId", bool useFeatureWeightFilter = false, int? numberOfExamplesToUse = default, int permutationCount = 1) where TModel : class;" />
<MemberSignature Language="ILAsm" Value=".method public static hidebysig valuetype System.Collections.Immutable.ImmutableArray`1<class Microsoft.ML.Data.RankingMetricsStatistics> PermutationFeatureImportance<class TModel>(class Microsoft.ML.RankingCatalog catalog, class Microsoft.ML.ISingleFeaturePredictionTransformer`1<!!TModel> predictionTransformer, class Microsoft.ML.IDataView data, string labelColumnName, string rowGroupColumnName, bool useFeatureWeightFilter, valuetype System.Nullable`1<int32> numberOfExamplesToUse, int32 permutationCount) cil managed" />
<MemberSignature Language="DocId" Value="M:Microsoft.ML.PermutationFeatureImportanceExtensions.PermutationFeatureImportance``1(Microsoft.ML.RankingCatalog,Microsoft.ML.ISingleFeaturePredictionTransformer{``0},Microsoft.ML.IDataView,System.String,System.String,System.Boolean,System.Nullable{System.Int32},System.Int32)" />
<MemberSignature Language="VB.NET" Value="<Extension()>
Public Function PermutationFeatureImportance(Of TModel As Class) (catalog As RankingCatalog, predictionTransformer As ISingleFeaturePredictionTransformer(Of TModel), data As IDataView, Optional labelColumnName As String = "Label", Optional rowGroupColumnName As String = "GroupId", Optional useFeatureWeightFilter As Boolean = false, Optional numberOfExamplesToUse As Nullable(Of Integer) = Nothing, Optional permutationCount As Integer = 1) As ImmutableArray(Of RankingMetricsStatistics)" />
<MemberSignature Language="F#" Value="static member PermutationFeatureImportance : Microsoft.ML.RankingCatalog * Microsoft.ML.ISingleFeaturePredictionTransformer<'Model (requires 'Model : null)> * Microsoft.ML.IDataView * string * string * bool * Nullable<int> * int -> System.Collections.Immutable.ImmutableArray<Microsoft.ML.Data.RankingMetricsStatistics> (requires 'Model : null)" Usage="Microsoft.ML.PermutationFeatureImportanceExtensions.PermutationFeatureImportance (catalog, predictionTransformer, data, labelColumnName, rowGroupColumnName, useFeatureWeightFilter, numberOfExamplesToUse, permutationCount)" />
<MemberType>Method</MemberType>
<AssemblyInfo>
<AssemblyName>Microsoft.ML.Transforms</AssemblyName>
<AssemblyVersion>1.0.0.0</AssemblyVersion>
</AssemblyInfo>
<ReturnValue>
<ReturnType>System.Collections.Immutable.ImmutableArray<Microsoft.ML.Data.RankingMetricsStatistics></ReturnType>
</ReturnValue>
<TypeParameters>
<TypeParameter Name="TModel">
<Constraints>
<ParameterAttribute>ReferenceTypeConstraint</ParameterAttribute>
</Constraints>
</TypeParameter>
</TypeParameters>
<Parameters>
<Parameter Name="catalog" Type="Microsoft.ML.RankingCatalog" RefType="this" />
<Parameter Name="predictionTransformer" Type="Microsoft.ML.ISingleFeaturePredictionTransformer<TModel>" />
<Parameter Name="data" Type="Microsoft.ML.IDataView" />
<Parameter Name="labelColumnName" Type="System.String" />
<Parameter Name="rowGroupColumnName" Type="System.String" />
<Parameter Name="useFeatureWeightFilter" Type="System.Boolean" />
<Parameter Name="numberOfExamplesToUse" Type="System.Nullable<System.Int32>" />
<Parameter Name="permutationCount" Type="System.Int32" />
</Parameters>
<Docs>
<typeparam name="TModel">To be added.</typeparam>
<param name="catalog">The ranking catalog.</param>
<param name="predictionTransformer">The model on which to evaluate feature importance.</param>
<param name="data">The evaluation data set.</param>
<param name="labelColumnName">Label column name. The column data must be <see cref="T:System.Single" /> or <see cref="T:Microsoft.ML.Data.KeyDataViewType" />.</param>
<param name="rowGroupColumnName">GroupId column name</param>
<param name="useFeatureWeightFilter">Use features weight to pre-filter features.</param>
<param name="numberOfExamplesToUse">Limit the number of examples to evaluate on. <cref langword="null" /> means up to ~2 bln examples from <paramref param="data" /> will be used.</param>
<param name="permutationCount">The number of permutations to perform.</param>
<summary>
Permutation Feature Importance (PFI) for Ranking.
</summary>
<returns>Array of per-feature 'contributions' to the score.</returns>
<remarks>
<para>
Permutation feature importance (PFI) is a technique to determine the global importance of features in a trained
machine learning model. PFI is a simple yet powerful technique motivated by Breiman in his Random Forest paper, section 10
(Breiman. <a href="https://www.stat.berkeley.edu/~breiman/randomforest2001.pdf">"Random Forests."</a> Machine Learning, 2001.)
The advantage of the PFI method is that it is model agnostic -- it works with any model that can be
evaluated -- and it can use any dataset, not just the training set, to compute feature importance metrics.
</para>
<para>
PFI works by taking a labeled dataset, choosing a feature, and permuting the values
for that feature across all the examples, so that each example now has a random value for the feature and
the original values for all other features. The evaluation metric (e.g. NDCG) is then calculated
for this modified dataset, and the change in the evaluation metric from the original dataset is computed.
The larger the change in the evaluation metric, the more important the feature is to the model.
PFI works by performing this permutation analysis across all the features of a model, one after another.
</para>
<para>
In this implementation, PFI computes the change in all possible ranking evaluation metrics for each feature, and an
<see cref="T:System.Collections.Immutable.ImmutableArray" /> of <see cref="T:Microsoft.ML.Data.RankingMetrics" /> objects is returned. See the sample below for an
example of working with these results to analyze the feature importance of a model.
</para>
</remarks>
<example>
<format type="text/markdown"><![CDATA[
[!code-csharp[PermutationFeatureImportance](~/../docs/samples/docs/samples/Microsoft.ML.Samples/Dynamic/Trainers/Ranking/PermutationFeatureImportance.cs)]
]]></format>
</example>
</Docs>
</Member>
<Member MemberName="PermutationFeatureImportanceNonCalibrated">
<MemberSignature Language="C#" Value="public static System.Collections.Immutable.ImmutableDictionary<string,Microsoft.ML.Data.BinaryClassificationMetricsStatistics> PermutationFeatureImportanceNonCalibrated (this Microsoft.ML.BinaryClassificationCatalog catalog, Microsoft.ML.ITransformer model, Microsoft.ML.IDataView data, string labelColumnName = "Label", bool useFeatureWeightFilter = false, int? numberOfExamplesToUse = default, int permutationCount = 1);" />
<MemberSignature Language="ILAsm" Value=".method public static hidebysig class System.Collections.Immutable.ImmutableDictionary`2<string, class Microsoft.ML.Data.BinaryClassificationMetricsStatistics> PermutationFeatureImportanceNonCalibrated(class Microsoft.ML.BinaryClassificationCatalog catalog, class Microsoft.ML.ITransformer model, class Microsoft.ML.IDataView data, string labelColumnName, bool useFeatureWeightFilter, valuetype System.Nullable`1<int32> numberOfExamplesToUse, int32 permutationCount) cil managed" />
<MemberSignature Language="DocId" Value="M:Microsoft.ML.PermutationFeatureImportanceExtensions.PermutationFeatureImportanceNonCalibrated(Microsoft.ML.BinaryClassificationCatalog,Microsoft.ML.ITransformer,Microsoft.ML.IDataView,System.String,System.Boolean,System.Nullable{System.Int32},System.Int32)" />
<MemberSignature Language="VB.NET" Value="<Extension()>
Public Function PermutationFeatureImportanceNonCalibrated (catalog As BinaryClassificationCatalog, model As ITransformer, data As IDataView, Optional labelColumnName As String = "Label", Optional useFeatureWeightFilter As Boolean = false, Optional numberOfExamplesToUse As Nullable(Of Integer) = Nothing, Optional permutationCount As Integer = 1) As ImmutableDictionary(Of String, BinaryClassificationMetricsStatistics)" />
<MemberSignature Language="F#" Value="static member PermutationFeatureImportanceNonCalibrated : Microsoft.ML.BinaryClassificationCatalog * Microsoft.ML.ITransformer * Microsoft.ML.IDataView * string * bool * Nullable<int> * int -> System.Collections.Immutable.ImmutableDictionary<string, Microsoft.ML.Data.BinaryClassificationMetricsStatistics>" Usage="Microsoft.ML.PermutationFeatureImportanceExtensions.PermutationFeatureImportanceNonCalibrated (catalog, model, data, labelColumnName, useFeatureWeightFilter, numberOfExamplesToUse, permutationCount)" />
<MemberType>Method</MemberType>
<AssemblyInfo>
<AssemblyName>Microsoft.ML.Transforms</AssemblyName>
<AssemblyVersion>1.0.0.0</AssemblyVersion>
</AssemblyInfo>
<ReturnValue>
<ReturnType>System.Collections.Immutable.ImmutableDictionary<System.String,Microsoft.ML.Data.BinaryClassificationMetricsStatistics></ReturnType>
</ReturnValue>
<Parameters>
<Parameter Name="catalog" Type="Microsoft.ML.BinaryClassificationCatalog" RefType="this" Index="0" FrameworkAlternate="ml-dotnet;ml-dotnet-1.7.0;ml-dotnet-2.0.0" />
<Parameter Name="model" Type="Microsoft.ML.ITransformer" Index="1" FrameworkAlternate="ml-dotnet;ml-dotnet-1.7.0;ml-dotnet-2.0.0" />
<Parameter Name="data" Type="Microsoft.ML.IDataView" Index="2" FrameworkAlternate="ml-dotnet;ml-dotnet-1.7.0;ml-dotnet-2.0.0" />
<Parameter Name="labelColumnName" Type="System.String" Index="3" FrameworkAlternate="ml-dotnet;ml-dotnet-1.7.0;ml-dotnet-2.0.0" />
<Parameter Name="useFeatureWeightFilter" Type="System.Boolean" Index="4" FrameworkAlternate="ml-dotnet;ml-dotnet-1.7.0;ml-dotnet-2.0.0" />
<Parameter Name="numberOfExamplesToUse" Type="System.Nullable<System.Int32>" Index="5" FrameworkAlternate="ml-dotnet;ml-dotnet-1.7.0;ml-dotnet-2.0.0" />
<Parameter Name="permutationCount" Type="System.Int32" Index="6" FrameworkAlternate="ml-dotnet;ml-dotnet-1.7.0;ml-dotnet-2.0.0" />
</Parameters>
<Docs>
<param name="catalog">The binary classification catalog.</param>
<param name="model">The model on which to evaluate feature importance.</param>
<param name="data">The evaluation data set.</param>
<param name="labelColumnName">Label column name. The column data must be <see cref="T:System.Boolean" />.</param>
<param name="useFeatureWeightFilter">Use features weight to pre-filter features.</param>
<param name="numberOfExamplesToUse">Limit the number of examples to evaluate on. <cref langword="null" /> means up to ~2 bln examples from <paramref param="data" /> will be used.</param>
<param name="permutationCount">The number of permutations to perform.</param>
<summary>
Permutation Feature Importance (PFI) for Binary Classification.
</summary>
<returns>Dictionary mapping each feature to its per-feature 'contributions' to the score.</returns>
<remarks>
<para>
Permutation feature importance (PFI) is a technique to determine the global importance of features in a trained
machine learning model. PFI is a simple yet powerful technique motivated by Breiman in his Random Forest paper, section 10
(Breiman. <a href="https://www.stat.berkeley.edu/~breiman/randomforest2001.pdf">"Random Forests."</a> Machine Learning, 2001.)
The advantage of the PFI method is that it is model agnostic -- it works with any model that can be
evaluated -- and it can use any dataset, not just the training set, to compute feature importance metrics.
</para>
<para>
PFI works by taking a labeled dataset, choosing a feature, and permuting the values
for that feature across all the examples, so that each example now has a random value for the feature and
the original values for all other features. The evaluation metric (e.g. AUC) is then calculated
for this modified dataset, and the change in the evaluation metric from the original dataset is computed.
The larger the change in the evaluation metric, the more important the feature is to the model.
PFI works by performing this permutation analysis across all the features of a model, one after another.
</para>
<para>
In this implementation, PFI computes the change in all possible binary classification evaluation metrics for each feature, and an
<see cref="T:System.Collections.Immutable.ImmutableArray" /> of <see cref="T:Microsoft.ML.Data.BinaryClassificationMetrics" /> objects is returned. See the sample below for an
example of working with these results to analyze the feature importance of a model.
</para>
</remarks>
<example>
<format type="text/markdown"><![CDATA[
[!code-csharp[PermutationFeatureImportance](~/../docs/samples/docs/samples/Microsoft.ML.Samples/Dynamic/Trainers/BinaryClassification/PermutationFeatureImportance.cs)]
]]></format>
</example>
</Docs>
</Member>
</Members>
</Type>