This repository has been archived by the owner on Nov 19, 2020. It is now read-only.
-
Notifications
You must be signed in to change notification settings - Fork 2k
/
Imputation.cs
647 lines (573 loc) · 22.1 KB
/
Imputation.cs
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
// Accord Statistics Library
// The Accord.NET Framework
// http://accord-framework.net
//
// Copyright © César Souza, 2009-2017
// cesarsouza at gmail.com
//
// This library is free software; you can redistribute it and/or
// modify it under the terms of the GNU Lesser General Public
// License as published by the Free Software Foundation; either
// version 2.1 of the License, or (at your option) any later version.
//
// This library is distributed in the hope that it will be useful,
// but WITHOUT ANY WARRANTY; without even the implied warranty of
// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
// Lesser General Public License for more details.
//
// You should have received a copy of the GNU Lesser General Public
// License along with this library; if not, write to the Free Software
// Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
//
namespace Accord.Statistics.Filters
{
using System;
using System.Data;
using System.Runtime.Serialization;
using Accord.Math;
using Accord.MachineLearning;
using System.Collections.Generic;
using Accord.Compat;
/// <summary>
/// Strategies for missing value imputations.
/// </summary>
///
/// <seealso cref="Imputation"/>
/// <seealso cref="Imputation{T}"/>
///
public enum ImputationStrategy
{
/// <summary>
/// Uses a fixed-value to replace missing fields.
/// </summary>
///
FixedValue,
/// <summary>
/// Uses the mean value to replace missing fields.
/// </summary>
///
Mean,
/// <summary>
/// Uses the mode value to replace missing fields.
/// </summary>
///
Mode,
/// <summary>
/// Uses the median value to replace missing fields.
/// </summary>
///
Median
};
/// <summary>
/// Imputation filter for filling missing values.
/// </summary>
///
/// <example>
/// <code source="Unit Tests\Accord.Tests.Statistics\Filters\ImputationFilterTest.cs" region="doc_learn" />
/// </example>
///
[Serializable]
public class Imputation : Imputation<object>
{
/// <summary>
/// Creates a new Imputation filter.
/// </summary>
///
public Imputation()
: base() { }
/// <summary>
/// Creates a new Imputation filter.
/// </summary>
///
public Imputation(string[] names, object[][] data)
: base(names, data)
{
}
/// <summary>
/// Creates a new Imputation filter.
/// </summary>
///
public Imputation(object[][] data)
: base(data)
{
}
/// <summary>
/// Creates a new Imputation filter.
/// </summary>
///
public Imputation(params string[] columns)
: base(columns)
{
}
}
/// <summary>
/// Imputation filter for filling missing values.
/// </summary>
///
/// <example>
/// <code source="Unit Tests\Accord.Tests.Statistics\Filters\ImputationFilterTest.cs" region="doc_learn" />
/// </example>
///
[Serializable]
public class Imputation<T> : BaseFilter<Imputation<T>.Options, Imputation<T>>,
IAutoConfigurableFilter, ITransform<T[], T[]>,
IUnsupervisedLearning<Imputation<T>, T[], T[]>
{
/// <summary>
/// Gets the number of outputs generated by the model.
/// </summary>
///
/// <value>The number of outputs.</value>
///
public int NumberOfOutputs { get { return Columns.Count; } }
/// <summary>
/// Creates a new Imputation filter.
/// </summary>
///
public Imputation()
: base() { }
/// <summary>
/// Creates a new Imputation filter.
/// </summary>
///
public Imputation(T[][] data)
{
this.Learn(data);
}
/// <summary>
/// Creates a new Imputation filter.
/// </summary>
///
public Imputation(string[] columnNames, T[][] data)
{
foreach (String col in columnNames)
Columns.Add(new Options(col));
this.Learn(data);
}
/// <summary>
/// Creates a new Imputation filter.
/// </summary>
///
public Imputation(params string[] columnNames)
{
foreach (String col in columnNames)
Columns.Add(new Options(col));
}
/// <summary>
/// Applies the transformation to an input, producing an associated output.
/// </summary>
/// <param name="input">The input data to which the transformation should be applied.</param>
/// <returns>The output generated by applying this transformation to the given input.</returns>
public T[] Transform(T[] input)
{
return Transform(new[] { input })[0];
}
/// <summary>
/// Applies the transformation to a set of input vectors,
/// producing an associated set of output vectors.
/// </summary>
/// <param name="input">The input data to which
/// the transformation should be applied.</param>
/// <returns>The output generated by applying this
/// transformation to the given input.</returns>
public T[][] Transform(T[][] input)
{
return Transform(input, Jagged.CreateAs(input));
}
/// <summary>
/// Applies the transformation to a set of input vectors,
/// producing an associated set of output vectors.
/// </summary>
/// <param name="input">The input data to which
/// the transformation should be applied.</param>
/// <param name="result">The location to where to store the
/// result of this transformation.</param>
/// <returns>The output generated by applying this
/// transformation to the given input.</returns>
public T[][] Transform(T[][] input, T[][] result)
{
for (int i = 0; i < input.Length; i++)
{
for (int j = 0; j < input[i].Length; j++)
{
Options options = Columns[j];
if (options.ReplaceWith != null && options.IsMissingValue(input[i][j]))
{
result[i][j] = Columns[j].ReplaceWith;
}
else
{
result[i][j] = input[i][j];
}
}
}
return result;
}
#if !NETSTANDARD1_4
/// <summary>
/// Processes the current filter.
/// </summary>
///
protected override DataTable ProcessFilter(DataTable data)
{
// Copy the DataTable
DataTable result = data.Copy();
foreach (DataRow row in result.Rows)
{
foreach (Options options in Columns)
{
if (options.ReplaceWith != null && options.IsMissingValue(row[options.ColumnName]))
{
row[options.ColumnName] = options.ReplaceWith;
}
}
}
return result;
}
/// <summary>
/// Auto detects the filter options by analyzing a given <see cref="System.Data.DataTable"/>.
/// </summary>
///
public void Detect(DataTable data)
{
Learn(data);
}
/// <summary>
/// Learns a model that can map the given inputs to the desired outputs.
/// </summary>
///
/// <param name="x">The model inputs.</param>
/// <param name="weights">The weight of importance for each input sample.</param>
///
/// <returns>A model that has learned how to produce suitable outputs
/// given the input data <paramref name="x" />.</returns>
///
/// <exception cref="ArgumentException">weights</exception>
/// <exception cref="Exception">There are more predefined columns than columns in the data.</exception>
///
public Imputation<T> Learn(DataTable x, double[] weights = null)
{
if (weights != null)
throw new ArgumentException(Accord.Properties.Resources.NotSupportedWeights, "weights");
foreach (DataColumn col in x.Columns)
{
if (!this.Columns.Contains(col.ColumnName))
{
Columns.Add(new Options(col.ColumnName));
}
}
foreach (DataColumn col in x.Columns)
{
this.Columns[col.ColumnName].Learn(x, weights);
}
return this;
}
#endif
/// <summary>
/// Learns a model that can map the given inputs to the desired outputs.
/// </summary>
///
/// <param name="x">The model inputs.</param>
/// <param name="weights">The weight of importance for each input sample.</param>
///
/// <returns>A model that has learned how to produce suitable outputs
/// given the input data <paramref name="x" />.</returns>
///
/// <exception cref="ArgumentException">weights</exception>
/// <exception cref="Exception">There are more predefined columns than columns in the data.</exception>
///
public Imputation<T> Learn(T[][] x, double[] weights = null)
{
if (weights != null)
throw new ArgumentException(Accord.Properties.Resources.NotSupportedWeights, "weights");
for (int i = this.Columns.Count; i < x.Columns(); i++)
this.Columns.Add(new Options(i.ToString()));
if (this.Columns.Count != x.Columns())
throw new Exception("There are more predefined columns than columns in the data.");
for (int i = 0; i < Columns.Count; i++)
Columns[i].Learn(x.GetColumn(i), weights);
return this;
}
int ITransform.NumberOfOutputs
{
get { return NumberOfOutputs; }
set { throw new InvalidOperationException("This property is read-only."); }
}
/// <summary>
/// Options for the imputation filter.
/// </summary>
///
[Serializable]
public class Options : ColumnOptionsBase<Imputation<T>>, IAutoConfigurableColumn
{
[OptionalField]
private ImputationStrategy strategy;
/// <summary>
/// Gets or sets the imputation strategy
/// to use with this column.
/// </summary>
public ImputationStrategy Strategy
{
get { return strategy; }
set { strategy = value; }
}
/// <summary>
/// Missing value indicator.
/// </summary>
///
public T MissingValue { get; set; }
/// <summary>
/// Value to replace missing values with.
/// </summary>
///
public T ReplaceWith { get; set; }
/// <summary>
/// Constructs a new column option
/// for the Imputation filter.
/// </summary>
///
public Options(String name)
: base(name)
{
if (typeof(T) == typeof(Double))
this.MissingValue = (Double.NaN).To<T>();
else if (typeof(T) == typeof(Single))
this.MissingValue = (Single.NaN).To<T>();
else if (typeof(T) == typeof(int))
this.MissingValue = (-1).To<T>();
else
this.MissingValue = default(T);
this.ReplaceWith = default(T);
}
/// <summary>
/// Constructs a new column option
/// for the Imputation filter.
/// </summary>
///
public Options()
: this("New column") { }
#if !NETSTANDARD1_4
/// <summary>
/// Auto detects the column options by analyzing
/// a given <see cref="System.Data.DataColumn"/>.
/// </summary>
///
/// <param name="column">The column to analyze.</param>
///
public void Detect(DataColumn column)
{
Learn(column.To<T[]>());
}
/// <summary>
/// Learns a model that can map the given inputs to the desired outputs.
/// </summary>
///
/// <param name="x">The model inputs.</param>
/// <param name="weights">The weight of importance for each input sample.</param>
///
/// <returns>A model that has learned how to produce suitable outputs
/// given the input data <paramref name="x" />.</returns>
///
/// <exception cref="ArgumentException">weights</exception>
/// <exception cref="Exception">There are more predefined columns than columns in the data.</exception>
///
public Options Learn(DataTable x, double[] weights = null)
{
if (weights != null)
throw new ArgumentException(Accord.Properties.Resources.NotSupportedWeights, "weights");
return Learn(x.Columns[this.ColumnName]);
}
/// <summary>
/// Learns a model that can map the given inputs to the desired outputs.
/// </summary>
///
/// <param name="x">The model inputs.</param>
/// <param name="weights">The weight of importance for each input sample.</param>
///
/// <returns>A model that has learned how to produce suitable outputs
/// given the input data <paramref name="x" />.</returns>
///
/// <exception cref="ArgumentException">weights</exception>
/// <exception cref="Exception">There are more predefined columns than columns in the data.</exception>
///
private Options Learn(DataColumn x, double[] weights = null)
{
Type type = x.DataType;
SetDefaultMissingValue(type);
if (strategy == ImputationStrategy.FixedValue)
return this;
if (strategy == ImputationStrategy.Mode)
{
object[] values = filter<object>(x);
ReplaceWith = values.Mode().To<T>();
return this;
}
if (type == typeof(double))
{
double[] doubleColumn = filter<double>(x);
switch (Strategy)
{
case ImputationStrategy.Mean:
ReplaceWith = doubleColumn.Mean().To<T>();
break;
case ImputationStrategy.Median:
ReplaceWith = doubleColumn.Median().To<T>();
break;
}
}
else if (type == typeof(int))
{
int[] intColumn = filter<int>(x);
switch (Strategy)
{
case ImputationStrategy.Mean:
ReplaceWith = intColumn.Mean().To<T>();
break;
case ImputationStrategy.Median:
ReplaceWith = intColumn.Median().To<T>();
break;
}
}
else
{
throw new NotSupportedException("The imputation strategy {0} is not supported for values of type {1}".Format(strategy, typeof(T)));
}
return this;
}
#endif
/// <summary>
/// Learns a model that can map the given inputs to the desired outputs.
/// </summary>
///
/// <param name="x">The model inputs.</param>
/// <param name="weights">The weight of importance for each input sample.</param>
///
/// <returns>A model that has learned how to produce suitable outputs
/// given the input data <paramref name="x" />.</returns>
///
/// <exception cref="ArgumentException">weights</exception>
/// <exception cref="Exception">There are more predefined columns than columns in the data.</exception>
///
public Options Learn(T[] x, double[] weights = null)
{
if (weights != null)
throw new ArgumentException(Accord.Properties.Resources.NotSupportedWeights, "weights");
Type type = x[0].GetType();
SetDefaultMissingValue(type);
if (strategy == ImputationStrategy.FixedValue)
return this;
x = filter(x);
if (strategy == ImputationStrategy.Mode)
{
ReplaceWith = x.Mode();
return this;
}
if (!compute(x))
{
if (type == typeof(double))
compute(x.To<double[]>());
else if (type == typeof(int))
compute(x.To<int[]>());
else
throw new NotSupportedException("The imputation strategy {0} is not supported for values of type {1}".Format(strategy, typeof(T)));
}
return this;
}
private void SetDefaultMissingValue(Type type)
{
if (typeof(T) != type && this.MissingValue == null)
{
if (type == typeof(int))
{
this.MissingValue = (-1).To<T>();
}
else if (type == typeof(double))
{
this.MissingValue = (Double.NaN).To<T>();
}
else if (type == typeof(float))
{
this.MissingValue = (Single.NaN).To<T>();
}
else
{
#if NETSTANDARD1_4
this.MissingValue = default(T);
#else
this.MissingValue = type.GetDefaultValue().To<T>();
#endif
}
}
}
/// <summary>
/// Determines whether the given object denotes a missing value.
/// </summary>
///
public bool IsMissingValue(object value)
{
#if !NETSTANDARD1_4
return (value is DBNull || value == null || Object.Equals(this.MissingValue, value));
#else
return (value == null || Object.Equals(this.MissingValue, value));
#endif
}
private bool compute(object column)
{
double[] doubleColumn = column as double[];
if (doubleColumn != null)
{
switch (Strategy)
{
case ImputationStrategy.Mean:
ReplaceWith = doubleColumn.Mean().To<T>();
break;
case ImputationStrategy.Median:
ReplaceWith = doubleColumn.Median().To<T>();
break;
}
return true;
}
int[] intColumn = column as int[];
if (intColumn != null)
{
switch (Strategy)
{
case ImputationStrategy.Mean:
ReplaceWith = intColumn.Mean().To<T>();
break;
case ImputationStrategy.Median:
ReplaceWith = intColumn.Median().To<T>();
break;
}
return true;
}
return false;
}
#if !NETSTANDARD1_4
private TValue[] filter<TValue>(DataColumn column)
{
var m = new List<TValue>();
foreach (DataRow row in column.Table.Rows)
{
object value = row[column];
if (Object.Equals(value, this.MissingValue) || value is DBNull)
continue;
m.Add((TValue)System.Convert.ChangeType(value, typeof(TValue)));
}
return m.ToArray();
}
#endif
private TValue[] filter<TValue>(TValue[] column)
{
var m = new List<TValue>();
foreach (TValue value in column)
{
if (Object.Equals(value, this.MissingValue) || value is DBNull)
continue;
m.Add((TValue)System.Convert.ChangeType(value, typeof(TValue)));
}
return m.ToArray();
}
}
}
}