Skip to content

Transformers

cesnet_tszoo.utils.transformer.transformer

Transformer

Bases: ABC

Base class for transformers, used for transforming data.

This class serves as the foundation for creating custom transformers. To implement a custom transformer, this class is recommended to be subclassed and extended.

Example:

import numpy as np

class LogTransformer(Transformer):

    def __init__(self):
        self.names = None

    def fit(self, data: np.ndarray) -> None:
        self.partial_fit(data)

    def partial_fit(self, data: np.ndarray) -> None:
        self.names = data.dtype.names

    def transform(self, data: np.ndarray) -> np.ndarray:

        for name in data.dtype.names:

            current_data = data[name]
            log_data = np.ma.log(current_data)
            current_data[:] = log_data.filled(np.nan)

        return data

    def inverse_transform(self, transformed_data: np.ndarray) -> np.ndarray:

        names = transformed_data.dtype.names if transformed_data.dtype.names is not None else self.names

        for name in names:

            current_data = transformed_data[name] if transformed_data.dtype.names is not None else transformed_data
            current_data[:] = np.exp(current_data)

        return transformed_data
Source code in cesnet_tszoo\utils\transformer\transformer.py 
  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
class Transformer(ABC):
    """
    Base class for transformers, used for transforming data.

    This class serves as the foundation for creating custom transformers. To implement a custom transformer, this class is recommended to be subclassed and extended.

    Example:

        import numpy as np

        class LogTransformer(Transformer):

            def __init__(self):
                self.names = None

            def fit(self, data: np.ndarray) -> None:
                self.partial_fit(data)

            def partial_fit(self, data: np.ndarray) -> None:
                self.names = data.dtype.names

            def transform(self, data: np.ndarray) -> np.ndarray:

                for name in data.dtype.names:

                    current_data = data[name]
                    log_data = np.ma.log(current_data)
                    current_data[:] = log_data.filled(np.nan)

                return data

            def inverse_transform(self, transformed_data: np.ndarray) -> np.ndarray:

                names = transformed_data.dtype.names if transformed_data.dtype.names is not None else self.names

                for name in names:

                    current_data = transformed_data[name] if transformed_data.dtype.names is not None else transformed_data
                    current_data[:] = np.exp(current_data)

                return transformed_data               
    """

    @abstractmethod
    def fit(self, data: np.ndarray) -> None:
        """
        Sets the transformer values for a given time series part.

        This method must be implemented if using multiple transformers that have not been pre-fitted.

        Parameters:
            data: A structured numpy array representing data for a single time series with shape `(times)`. Use data["base_data"] to get non matrix features excluding any identifiers. 
                  For matrix features use their name instead of base_data.
        """
        ...

    @abstractmethod
    def partial_fit(self, data: np.ndarray) -> None:
        """
        Partially sets the transformer values for a given time series part.

        This method must be implemented if using a single transformer that is not pre-fitted for all time series, or when using pre-fitted transformer(s) with `partial_fit_initialized_transformers` set to `True`.

        Parameters:
            data: A structured numpy array representing data for a single time series with shape `(times)`. Use data["base_data"] to get non matrix features excluding any identifiers. 
                  For matrix features use their name instead of base_data.   
        """
        ...

    @abstractmethod
    def transform(self, data: np.ndarray) -> np.ndarray:
        """
        Transforms the input data for a given time series part.

        This method must always be implemented.

        Parameters:
            data: A structured numpy array representing data for a single time series with shape `(times)`. Use data["base_data"] to get non matrix features excluding any identifiers. 
                  For matrix features use their name instead of base_data.

        Returns:
            The transformed data, with the same shape and dtype as the input `(times)`.            
        """
        ...

    def inverse_transform(self, transformed_data: np.ndarray) -> np.ndarray:
        """
        Transforms the input transformed data to their original representation for a given time series part.

        Parameters:
            transformed_data: A structured numpy array representing data for a single time series with shape `(times)`. Use data["base_data"] to get non matrix features excluding any identifiers. 
                              For matrix features use their name instead of base_data. 

        Returns:
            The original representation of transformed data, with the same shape and dtype as the input `(times)`.            
        """
        return transformed_data

fit abstractmethod 

fit(data: ndarray) -> None

Sets the transformer values for a given time series part.

This method must be implemented if using multiple transformers that have not been pre-fitted.

Parameters:

Name Type Description Default
data ndarray

A structured numpy array representing data for a single time series with shape (times). Use data["base_data"] to get non matrix features excluding any identifiers. For matrix features use their name instead of base_data.

 required
Source code in cesnet_tszoo\utils\transformer\transformer.py 
52
53
54
55
56
57
58
59
60
61
62
63
@abstractmethod
def fit(self, data: np.ndarray) -> None:
    """
    Sets the transformer values for a given time series part.

    This method must be implemented if using multiple transformers that have not been pre-fitted.

    Parameters:
        data: A structured numpy array representing data for a single time series with shape `(times)`. Use data["base_data"] to get non matrix features excluding any identifiers. 
              For matrix features use their name instead of base_data.
    """
    ...

inverse_transform 

inverse_transform(transformed_data: ndarray) -> np.ndarray

Transforms the input transformed data to their original representation for a given time series part.

Parameters:

Name Type Description Default
transformed_data ndarray

A structured numpy array representing data for a single time series with shape (times). Use data["base_data"] to get non matrix features excluding any identifiers. For matrix features use their name instead of base_data.

required

Returns:

Type Description
ndarray

The original representation of transformed data, with the same shape and dtype as the input (times).
Source code in cesnet_tszoo\utils\transformer\transformer.py 
 94
 95
 96
 97
 98
 99
100
101
102
103
104
105
def inverse_transform(self, transformed_data: np.ndarray) -> np.ndarray:
    """
    Transforms the input transformed data to their original representation for a given time series part.

    Parameters:
        transformed_data: A structured numpy array representing data for a single time series with shape `(times)`. Use data["base_data"] to get non matrix features excluding any identifiers. 
                          For matrix features use their name instead of base_data. 

    Returns:
        The original representation of transformed data, with the same shape and dtype as the input `(times)`.            
    """
    return transformed_data

partial_fit abstractmethod 

partial_fit(data: ndarray) -> None

Partially sets the transformer values for a given time series part.

This method must be implemented if using a single transformer that is not pre-fitted for all time series, or when using pre-fitted transformer(s) with partial_fit_initialized_transformers set to True.

Parameters:

Name Type Description Default
data ndarray

A structured numpy array representing data for a single time series with shape (times). Use data["base_data"] to get non matrix features excluding any identifiers. For matrix features use their name instead of base_data.

 required
Source code in cesnet_tszoo\utils\transformer\transformer.py 
65
66
67
68
69
70
71
72
73
74
75
76
@abstractmethod
def partial_fit(self, data: np.ndarray) -> None:
    """
    Partially sets the transformer values for a given time series part.

    This method must be implemented if using a single transformer that is not pre-fitted for all time series, or when using pre-fitted transformer(s) with `partial_fit_initialized_transformers` set to `True`.

    Parameters:
        data: A structured numpy array representing data for a single time series with shape `(times)`. Use data["base_data"] to get non matrix features excluding any identifiers. 
              For matrix features use their name instead of base_data.   
    """
    ...

transform abstractmethod 

transform(data: ndarray) -> np.ndarray

Transforms the input data for a given time series part.

This method must always be implemented.

Parameters:

Name Type Description Default
data ndarray

A structured numpy array representing data for a single time series with shape (times). Use data["base_data"] to get non matrix features excluding any identifiers. For matrix features use their name instead of base_data.

 required

Returns:

Type Description
ndarray

The transformed data, with the same shape and dtype as the input (times).
Source code in cesnet_tszoo\utils\transformer\transformer.py 
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
@abstractmethod
def transform(self, data: np.ndarray) -> np.ndarray:
    """
    Transforms the input data for a given time series part.

    This method must always be implemented.

    Parameters:
        data: A structured numpy array representing data for a single time series with shape `(times)`. Use data["base_data"] to get non matrix features excluding any identifiers. 
              For matrix features use their name instead of base_data.

    Returns:
        The transformed data, with the same shape and dtype as the input `(times)`.            
    """
    ...

LogTransformer

Bases: Transformer

Tranforms data with natural logarithm. Zero or invalid values are set to np.nan.

Corresponds to enum TransformerType.LOG_TRANSFORMER or literal log_transformer.

Source code in cesnet_tszoo\utils\transformer\transformer.py 
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
class LogTransformer(Transformer):
    """
    Tranforms data with natural logarithm. Zero or invalid values are set to `np.nan`.

    Corresponds to enum [`TransformerType.LOG_TRANSFORMER`][cesnet_tszoo.utils.enums.TransformerType] or literal `log_transformer`.
    """

    def __init__(self):
        self.names = None

    def fit(self, data: np.ndarray) -> None:
        self.partial_fit(data)

    def partial_fit(self, data: np.ndarray) -> None:
        self.names = data.dtype.names

    def transform(self, data: np.ndarray) -> np.ndarray:

        for name in data.dtype.names:

            current_data = data[name]
            log_data = np.ma.log(current_data)
            current_data[:] = log_data.filled(np.nan)

        return data

    def inverse_transform(self, transformed_data: np.ndarray) -> np.ndarray:

        names = transformed_data.dtype.names if transformed_data.dtype.names is not None else self.names

        for name in names:

            current_data = transformed_data[name] if transformed_data.dtype.names is not None else transformed_data
            current_data[:] = np.exp(current_data)

        return transformed_data

L2Normalizer

Bases: Transformer

Tranforms data using Scikit L2Normalizer.

Corresponds to enum TransformerType.L2_NORMALIZER or literal l2_normalizer.

Source code in cesnet_tszoo\utils\transformer\transformer.py 
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
class L2Normalizer(Transformer):
    """
    Tranforms data using Scikit [`L2Normalizer`](https://scikit-learn.org/stable/modules/generated/sklearn.preprocessing.Normalizer.html).

    Corresponds to enum [`TransformerType.L2_NORMALIZER`][cesnet_tszoo.utils.enums.TransformerType] or literal `l2_normalizer`.
    """

    def __init__(self):
        self.transformer = sk.Normalizer(norm="l2")

    def fit(self, data: np.ndarray) -> None:
        ...

    def partial_fit(self, data: np.ndarray) -> None:
        ...

    def transform(self, data: np.ndarray) -> np.ndarray:
        for name in data.dtype.names:
            current_data = data[name]
            original_shape = current_data.shape
            flat_size = int(np.prod(original_shape[1:]))

            current_data = current_data.reshape(current_data.shape[0], flat_size)
            current_data[:] = self.transformer.transform(current_data)

        return data

    def inverse_transform(self, transformed_data: np.ndarray) -> np.ndarray:
        raise NotImplementedError("Normalizer does not support inverse_transform.")

MinMaxScaler

Bases: Transformer

Tranforms data using Scikit MinMaxScaler.

Corresponds to enum TransformerType.MIN_MAX_SCALER or literal min_max_scaler.

Source code in cesnet_tszoo\utils\transformer\transformer.py 
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
class MinMaxScaler(Transformer):
    """
    Tranforms data using Scikit [`MinMaxScaler`](https://scikit-learn.org/stable/modules/generated/sklearn.preprocessing.MinMaxScaler.html).

    Corresponds to enum [`TransformerType.MIN_MAX_SCALER`][cesnet_tszoo.utils.enums.TransformerType] or literal `min_max_scaler`.
    """

    def __init__(self):
        self.transformers: dict[str, sk.MinMaxScaler] = {}

    def fit(self, data: np.ndarray) -> None:
        self.partial_fit(data)

    def partial_fit(self, data: np.ndarray) -> None:

        is_init = len(self.transformers) == 0

        for name in data.dtype.names:

            if is_init:
                transformer = self.transformers[name] = sk.MinMaxScaler()
            else:
                transformer = self.transformers[name]

            current_data = data[name]
            flat_size = int(np.prod(current_data.shape[1:]))

            current_data = current_data.reshape(current_data.shape[0], flat_size)

            transformer.partial_fit(current_data)

    def transform(self, data: np.ndarray) -> np.ndarray:

        for name in data.dtype.names:

            transformer = self.transformers[name]
            current_data = data[name]
            original_shape = current_data.shape
            flat_size = int(np.prod(original_shape[1:]))

            current_data = current_data.reshape(current_data.shape[0], flat_size)

            current_data[:] = transformer.transform(current_data)

        return data

    def inverse_transform(self, transformed_data: np.ndarray) -> np.ndarray:

        names = transformed_data.dtype.names if transformed_data.dtype.names is not None else self.transformers.keys()

        for name in names:

            transformer = self.transformers[name]
            current_data = transformed_data[name] if transformed_data.dtype.names is not None else transformed_data
            original_shape = current_data.shape
            flat_size = int(np.prod(original_shape[1:]))

            current_data = current_data.reshape(current_data.shape[0], flat_size)

            current_data[:] = transformer.inverse_transform(current_data)

        return transformed_data

StandardScaler

Bases: Transformer

Tranforms data using Scikit StandardScaler.

Corresponds to enum TransformerType.STANDARD_SCALER or literal standard_scaler.

Source code in cesnet_tszoo\utils\transformer\transformer.py 
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
class StandardScaler(Transformer):
    """
    Tranforms data using Scikit [`StandardScaler`](https://scikit-learn.org/stable/modules/generated/sklearn.preprocessing.StandardScaler.html).

    Corresponds to enum [`TransformerType.STANDARD_SCALER`][cesnet_tszoo.utils.enums.TransformerType] or literal `standard_scaler`.
    """

    def __init__(self):
        self.transformers: dict[str, sk.StandardScaler] = {}

    def fit(self, data: np.ndarray) -> None:
        self.partial_fit(data)

    def partial_fit(self, data: np.ndarray) -> None:

        is_init = len(self.transformers) == 0

        for name in data.dtype.names:

            if is_init:
                transformer = self.transformers[name] = sk.StandardScaler()
            else:
                transformer = self.transformers[name]

            current_data = data[name]
            flat_size = int(np.prod(current_data.shape[1:]))

            current_data = current_data.reshape(current_data.shape[0], flat_size)

            transformer.partial_fit(current_data)

    def transform(self, data: np.ndarray) -> np.ndarray:

        for name in data.dtype.names:

            transformer = self.transformers[name]
            current_data = data[name]
            original_shape = current_data.shape
            flat_size = int(np.prod(original_shape[1:]))

            current_data = current_data.reshape(current_data.shape[0], flat_size)

            current_data[:] = transformer.transform(current_data)

        return data

    def inverse_transform(self, transformed_data: np.ndarray) -> np.ndarray:

        names = transformed_data.dtype.names if transformed_data.dtype.names is not None else self.transformers.keys()

        for name in names:

            transformer = self.transformers[name]
            current_data = transformed_data[name] if transformed_data.dtype.names is not None else transformed_data
            original_shape = current_data.shape
            flat_size = int(np.prod(original_shape[1:]))

            current_data = current_data.reshape(current_data.shape[0], flat_size)

            current_data[:] = transformer.inverse_transform(current_data)

        return transformed_data

MaxAbsScaler

Bases: Transformer

Tranforms data using Scikit MaxAbsScaler.

Corresponds to enum TransformerType.MAX_ABS_SCALER or literal max_abs_scaler.

Source code in cesnet_tszoo\utils\transformer\transformer.py 
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
class MaxAbsScaler(Transformer):
    """
    Tranforms data using Scikit [`MaxAbsScaler`](https://scikit-learn.org/stable/modules/generated/sklearn.preprocessing.MaxAbsScaler.html).

    Corresponds to enum [`TransformerType.MAX_ABS_SCALER`][cesnet_tszoo.utils.enums.TransformerType] or literal `max_abs_scaler`.
    """

    def __init__(self):
        self.transformers: dict[str, sk.MaxAbsScaler] = {}

    def fit(self, data: np.ndarray) -> None:
        self.partial_fit(data)

    def partial_fit(self, data: np.ndarray) -> None:

        is_init = len(self.transformers) == 0

        for name in data.dtype.names:

            if is_init:
                transformer = self.transformers[name] = sk.MaxAbsScaler()
            else:
                transformer = self.transformers[name]

            current_data = data[name]
            flat_size = int(np.prod(current_data.shape[1:]))

            current_data = current_data.reshape(current_data.shape[0], flat_size)

            transformer.partial_fit(current_data)

    def transform(self, data: np.ndarray) -> np.ndarray:

        for name in data.dtype.names:

            transformer = self.transformers[name]
            current_data = data[name]
            original_shape = current_data.shape
            flat_size = int(np.prod(original_shape[1:]))

            current_data = current_data.reshape(current_data.shape[0], flat_size)

            current_data[:] = transformer.transform(current_data)

        return data

    def inverse_transform(self, transformed_data: np.ndarray) -> np.ndarray:

        names = transformed_data.dtype.names if transformed_data.dtype.names is not None else self.transformers.keys()

        for name in names:

            transformer = self.transformers[name]
            current_data = transformed_data[name] if transformed_data.dtype.names is not None else transformed_data
            original_shape = current_data.shape
            flat_size = int(np.prod(original_shape[1:]))

            current_data = current_data.reshape(current_data.shape[0], flat_size)

            current_data[:] = transformer.inverse_transform(current_data)

        return transformed_data

PowerTransformer

Bases: Transformer

Tranforms data using Scikit PowerTransformer.

Corresponds to enum TransformerType.POWER_TRANSFORMER or literal power_transformer.

partial_fit not supported

Because this transformer does not support partial_fit it can't be used when using one transformer that needs to be fitted for multiple time series.

Source code in cesnet_tszoo\utils\transformer\transformer.py 
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
class PowerTransformer(Transformer):
    """
    Tranforms data using Scikit [`PowerTransformer`](https://scikit-learn.org/stable/modules/generated/sklearn.preprocessing.PowerTransformer.html).

    Corresponds to enum [`TransformerType.POWER_TRANSFORMER`][cesnet_tszoo.utils.enums.TransformerType] or literal `power_transformer`.

    !!! warning "partial_fit not supported"
        Because this transformer does not support partial_fit it can't be used when using one transformer that needs to be fitted for multiple time series.
    """

    def __init__(self):
        self.transformers: dict[str, sk.PowerTransformer] = {}

    def fit(self, data: np.ndarray) -> None:

        for name in data.dtype.names:
            transformer = self.transformers[name] = sk.PowerTransformer()
            current_data = data[name]
            flat_size = int(np.prod(current_data.shape[1:]))

            current_data = current_data.reshape(current_data.shape[0], flat_size)
            transformer.fit(current_data)

    def partial_fit(self, data: np.ndarray) -> None:
        raise NotImplementedError("PowerTransformer does not support partial_fit.")

    def transform(self, data: np.ndarray) -> np.ndarray:

        for name in data.dtype.names:
            transformer = self.transformers[name]
            current_data = data[name]
            original_shape = current_data.shape
            flat_size = int(np.prod(original_shape[1:]))

            current_data = current_data.reshape(current_data.shape[0], flat_size)
            current_data[:] = transformer.transform(current_data)

        return data

    def inverse_transform(self, transformed_data: np.ndarray) -> np.ndarray:

        names = transformed_data.dtype.names if transformed_data.dtype.names is not None else self.transformers.keys()

        for name in names:
            transformer = self.transformers[name]
            current_data = transformed_data[name] if transformed_data.dtype.names is not None else transformed_data
            original_shape = current_data.shape
            flat_size = int(np.prod(original_shape[1:]))

            current_data = current_data.reshape(current_data.shape[0], flat_size)
            current_data[:] = transformer.inverse_transform(current_data)

        return transformed_data

QuantileTransformer

Bases: Transformer

Tranforms data using Scikit QuantileTransformer.

Corresponds to enum TransformerType.QUANTILE_TRANSFORMER or literal quantile_transformer.

partial_fit not supported

Because this transformer does not support partial_fit it can't be used when using one transformer that needs to be fitted for multiple time series.

Source code in cesnet_tszoo\utils\transformer\transformer.py 
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
class QuantileTransformer(Transformer):
    """
    Tranforms data using Scikit [`QuantileTransformer`](https://scikit-learn.org/stable/modules/generated/sklearn.preprocessing.QuantileTransformer.html).

    Corresponds to enum [`TransformerType.QUANTILE_TRANSFORMER`][cesnet_tszoo.utils.enums.TransformerType] or literal `quantile_transformer`.

    !!! warning "partial_fit not supported"
        Because this transformer does not support partial_fit it can't be used when using one transformer that needs to be fitted for multiple time series.    
    """

    def __init__(self):
        self.transformers: dict[str, sk.QuantileTransformer] = {}

    def fit(self, data: np.ndarray) -> None:

        for name in data.dtype.names:
            transformer = self.transformers[name] = sk.QuantileTransformer()
            current_data = data[name]
            flat_size = int(np.prod(current_data.shape[1:]))

            current_data = current_data.reshape(current_data.shape[0], flat_size)
            transformer.fit(current_data)

    def partial_fit(self, data: np.ndarray) -> None:
        raise NotImplementedError("QuantileTransformer does not support partial_fit.")

    def transform(self, data: np.ndarray) -> np.ndarray:

        for name in data.dtype.names:
            transformer = self.transformers[name]
            current_data = data[name]
            original_shape = current_data.shape
            flat_size = int(np.prod(original_shape[1:]))

            current_data = current_data.reshape(current_data.shape[0], flat_size)
            current_data[:] = transformer.transform(current_data)

        return data

    def inverse_transform(self, transformed_data: np.ndarray) -> np.ndarray:

        names = transformed_data.dtype.names if transformed_data.dtype.names is not None else self.transformers.keys()

        for name in names:
            transformer = self.transformers[name]
            current_data = transformed_data[name] if transformed_data.dtype.names is not None else transformed_data
            original_shape = current_data.shape
            flat_size = int(np.prod(original_shape[1:]))

            current_data = current_data.reshape(current_data.shape[0], flat_size)
            current_data[:] = transformer.inverse_transform(current_data)

        return transformed_data

RobustScaler

Bases: Transformer

Tranforms data using Scikit RobustScaler.

Corresponds to enum TransformerType.ROBUST_SCALER or literal robust_scaler.

partial_fit not supported

Because this transformer does not support partial_fit it can't be used when using one transformer that needs to be fitted for multiple time series.

Source code in cesnet_tszoo\utils\transformer\transformer.py 
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
class RobustScaler(Transformer):
    """
    Tranforms data using Scikit [`RobustScaler`](https://scikit-learn.org/stable/modules/generated/sklearn.preprocessing.RobustScaler.html).

    Corresponds to enum [`TransformerType.ROBUST_SCALER`][cesnet_tszoo.utils.enums.TransformerType] or literal `robust_scaler`.

    !!! warning "partial_fit not supported"
        Because this transformer does not support partial_fit it can't be used when using one transformer that needs to be fitted for multiple time series.    
    """

    def __init__(self):
        self.transformers: dict[str, sk.RobustScaler] = {}

    def fit(self, data: np.ndarray) -> None:

        for name in data.dtype.names:
            transformer = self.transformers[name] = sk.RobustScaler()
            current_data = data[name]
            flat_size = int(np.prod(current_data.shape[1:]))

            current_data = current_data.reshape(current_data.shape[0], flat_size)
            transformer.fit(current_data)

    def partial_fit(self, data: np.ndarray) -> None:
        raise NotImplementedError("RobustScaler does not support partial_fit.")

    def transform(self, data: np.ndarray) -> np.ndarray:

        for name in data.dtype.names:
            transformer = self.transformers[name]
            current_data = data[name]
            original_shape = current_data.shape
            flat_size = int(np.prod(original_shape[1:]))

            current_data = current_data.reshape(current_data.shape[0], flat_size)
            current_data[:] = transformer.transform(current_data)

        return data

    def inverse_transform(self, transformed_data: np.ndarray) -> np.ndarray:

        names = transformed_data.dtype.names if transformed_data.dtype.names is not None else self.transformers.keys()

        for name in names:
            transformer = self.transformers[name]
            current_data = transformed_data[name] if transformed_data.dtype.names is not None else transformed_data
            original_shape = current_data.shape
            flat_size = int(np.prod(original_shape[1:]))

            current_data = current_data.reshape(current_data.shape[0], flat_size)
            current_data[:] = transformer.inverse_transform(current_data)

        return transformed_data

NoTransformer

Bases: Transformer

Does nothing.

Corresponds to enum TransformerType.NO_TRANSFORMER or literal no_transformer.

Source code in cesnet_tszoo\utils\transformer\transformer.py 
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
class NoTransformer(Transformer):
    """
    Does nothing.

    Corresponds to enum [`TransformerType.NO_TRANSFORMER`][cesnet_tszoo.utils.enums.TransformerType] or literal `no_transformer`.
    """

    def fit(self, data: np.ndarray):
        ...

    def partial_fit(self, data: np.ndarray) -> None:
        ...

    def transform(self, data: np.ndarray) -> np.ndarray:
        return data

    def inverse_transform(self, transformed_data: np.ndarray) -> np.ndarray:
        return transformed_data