Skip to content

ForecasterRnn

skforecast.deep_learning.utils.create_and_compile_model 

create_and_compile_model(
    series,
    lags,
    steps,
    levels=None,
    exog=None,
    recurrent_layer="LSTM",
    recurrent_units=100,
    recurrent_layers_kwargs={"activation": "tanh"},
    dense_units=64,
    dense_layers_kwargs={"activation": "relu"},
    output_dense_layer_kwargs={"activation": "linear"},
    compile_kwargs={
        "optimizer": Adam(),
        "loss": MeanSquaredError(),
    },
    model_name=None,
)

Build and compile a RNN-based Keras model for time series prediction, supporting exogenous variables.

Parameters:

Name Type Description Default
series pandas DataFrame

Input time series with shape (n_obs, n_series). Each column is a time series.

 required
lags int, list, numpy ndarray, range

Number of lagged time steps to consider in the input, index starts at 1, so lag 1 is equal to t-1.

  • int: include lags from 1 to lags (included).
  • list, 1d numpy ndarray or range: include only lags present in lags, all elements must be int.
 required
steps int

Number of steps to predict.

 required
levels (str, list)

Output level(s) (features) to predict. If None, defaults to the names of input series.

 None
exog pandas Series, pandas DataFrame

Exogenous variables to be included as input, should have the same number of rows as series.

 None
recurrent_layer str

Type of recurrent layer to be used, 'LSTM' [1], 'GRU' [2], or 'RNN' [3]_.

 'LSTM'
recurrent_units (int, list)

Number of units in the recurrent layer(s). Can be an integer for single recurrent layer, or a list of integers for multiple recurrent layers.

 100
recurrent_layers_kwargs (dict, list)

Additional keyword arguments for the recurrent layers [1], [2], [3]_. Can be a single dictionary for all layers or a list of dictionaries specifying different parameters for each recurrent layer.

 {'activation': 'tanh'}
dense_units (int, list, tuple, None)

Number of units in the dense layer(s) [4]_. Can be an integer for single dense layer, or a list of integers for multiple dense layers.

 64
dense_layers_kwargs (dict, list)

Additional keyword arguments for the dense layers [4]_. Can be a single dictionary for all layers or a list of dictionaries specifying different parameters for each dense layer.

 {'activation': 'relu'}
output_dense_layer_kwargs dict

Additional keyword arguments for the output dense layer.

 {'activation': 'linear'}
compile_kwargs dict

Additional keyword arguments for the model compilation, such as optimizer and loss function. [5]_

 {'optimizer': Adam(), 'loss': MeanSquaredError()}
model_name str

Name of the model.

 None

Returns:

Name Type Description
model Model

Compiled Keras model ready for training.
References

.. [1] LSTM layer Keras documentation. https://keras.io/api/layers/recurrent_layers/lstm/

.. [2] GRU layer Keras documentation. https://keras.io/api/layers/recurrent_layers/gru/

.. [3] SimpleRNN layer Keras documentation. https://keras.io/api/layers/recurrent_layers/simple_rnn/

.. [4] Dense layer Keras documentation. https://keras.io/api/layers/core_layers/dense/

.. [5] Model training APIs: compile method. https://keras.io/api/models/model_training_apis/

Source code in skforecast/deep_learning/utils.py 
 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
def create_and_compile_model(
    series: pd.DataFrame,
    lags: int | list[int] | np.ndarray[int] | range[int],
    steps: int,
    levels: str | list[str] | tuple[str] | None = None,
    exog: pd.Series | pd.DataFrame | None = None,
    recurrent_layer: str = "LSTM",
    recurrent_units: int | list[int] | tuple[int] = 100,
    recurrent_layers_kwargs: dict[str, Any] | list[dict[str, Any]] | None = {"activation": "tanh"},
    dense_units: int | list[int] | tuple[int] | None = 64,
    dense_layers_kwargs: dict[str, Any] | list[dict[str, Any]] | None = {"activation": "relu"},
    output_dense_layer_kwargs: dict[str, Any] | None = {"activation": "linear"},
    compile_kwargs: dict[str, Any] = {"optimizer": Adam(), "loss": MeanSquaredError()},
    model_name: str | None = None
) -> keras.models.Model:
    """
    Build and compile a RNN-based Keras model for time series prediction, 
    supporting exogenous variables.

    Parameters
    ----------
    series : pandas DataFrame
        Input time series with shape (n_obs, n_series). Each column is a time series.
    lags : int, list, numpy ndarray, range
        Number of lagged time steps to consider in the input, index starts at 1, 
        so lag 1 is equal to t-1.

        - `int`: include lags from 1 to `lags` (included).
        - `list`, `1d numpy ndarray` or `range`: include only lags present in 
        `lags`, all elements must be int.
    steps : int
        Number of steps to predict.
    levels : str, list, default None
        Output level(s) (features) to predict. If None, defaults to the names of 
        input series.
    exog : pandas Series, pandas DataFrame, default None
        Exogenous variables to be included as input, should have the same number 
        of rows as `series`.
    recurrent_layer : str, default 'LSTM'
        Type of recurrent layer to be used, 'LSTM' [1]_, 'GRU' [2]_, or 'RNN' [3]_.
    recurrent_units : int, list, default 100
        Number of units in the recurrent layer(s). Can be an integer for single 
        recurrent layer, or a list of integers for multiple recurrent layers.
    recurrent_layers_kwargs : dict, list, default {'activation': 'tanh'}
        Additional keyword arguments for the recurrent layers [1]_, [2]_, [3]_. 
        Can be a single dictionary for all layers or a list of dictionaries 
        specifying different parameters for each recurrent layer.
    dense_units : int, list, tuple, None, default 64
        Number of units in the dense layer(s) [4]_. Can be an integer for single
        dense layer, or a list of integers for multiple dense layers.
    dense_layers_kwargs : dict, list, default {'activation': 'relu'}
        Additional keyword arguments for the dense layers [4]_. Can be a single
        dictionary for all layers or a list of dictionaries specifying different
        parameters for each dense layer.
    output_dense_layer_kwargs : dict, default {'activation': 'linear'}
        Additional keyword arguments for the output dense layer.
    compile_kwargs : dict, default {'optimizer': Adam(), 'loss': MeanSquaredError()}
        Additional keyword arguments for the model compilation, such as optimizer 
        and loss function. [5]_
    model_name : str, default None
        Name of the model.

    Returns
    -------
    model : keras.models.Model
        Compiled Keras model ready for training.

    References
    ----------
    .. [1] LSTM layer Keras documentation.
           https://keras.io/api/layers/recurrent_layers/lstm/

    .. [2] GRU layer Keras documentation.
           https://keras.io/api/layers/recurrent_layers/gru/

    .. [3] SimpleRNN layer Keras documentation.
           https://keras.io/api/layers/recurrent_layers/simple_rnn/

    .. [4] Dense layer Keras documentation.
           https://keras.io/api/layers/core_layers/dense/

    .. [5] Model training APIs: compile method.
           https://keras.io/api/models/model_training_apis/

    """

    keras_backend = keras.backend.backend()

    print(f"keras version: {keras.__version__}")
    print(f"Using backend: {keras_backend}")
    if keras_backend == "tensorflow":
        import tensorflow
        print(f"tensorflow version: {tensorflow.__version__}")
    elif keras_backend == "torch":
        import torch
        print(f"torch version: {torch.__version__}")
    elif keras_backend == "jax":
        import jax
        print(f"jax version: {jax.__version__}")
    else:
        print("Backend not recognized")
    print("")

    if exog is None:
        model = _create_and_compile_model_no_exog(
            series=series,
            lags=lags,
            steps=steps,
            levels=levels,
            recurrent_layer=recurrent_layer,
            recurrent_units=recurrent_units,
            recurrent_layers_kwargs=recurrent_layers_kwargs,
            dense_units=dense_units,
            dense_layers_kwargs=dense_layers_kwargs,
            output_dense_layer_kwargs=output_dense_layer_kwargs,
            compile_kwargs=compile_kwargs,
            model_name=model_name
        )
    else:
        model = _create_and_compile_model_exog(
            series=series,
            lags=lags,
            steps=steps,
            levels=levels,
            exog=exog,
            recurrent_layer=recurrent_layer,
            recurrent_units=recurrent_units,
            recurrent_layers_kwargs=recurrent_layers_kwargs,
            dense_units=dense_units,
            dense_layers_kwargs=dense_layers_kwargs,
            output_dense_layer_kwargs=output_dense_layer_kwargs,
            compile_kwargs=compile_kwargs,
            model_name=model_name
        )

    return model

skforecast.deep_learning._forecaster_rnn.ForecasterRnn 

ForecasterRnn(
    levels,
    lags,
    estimator=None,
    transformer_series=MinMaxScaler(feature_range=(0, 1)),
    transformer_exog=MinMaxScaler(feature_range=(0, 1)),
    fit_kwargs={},
    forecaster_id=None,
    regressor=None,
)

Bases: ForecasterBase

This class turns any estimator compatible with the Keras API into a Keras RNN multi-series multi-step forecaster. A unique model is created to forecast all time steps and series. Keras enables workflows on top of either JAX, TensorFlow, or PyTorch. See documentation for more details.

Parameters:

Name Type Description Default
estimator estimator or pipeline compatible with the Keras API

An instance of a estimator or pipeline compatible with the Keras API.

 None
levels (str, list)

Name of one or more time series to be predicted. This determine the series the forecaster will be handling. If None, all series used during training will be available for prediction.

 required
lags int, list, numpy ndarray, range

Lags used as predictors. Index starts at 1, so lag 1 is equal to t-1.

  • int: include lags from 1 to lags (included).
  • list, 1d numpy ndarray or range: include only lags present in lags, all elements must be int.
 required
transformer_series (transformer(preprocessor), dict)

An instance of a transformer (preprocessor) compatible with the scikit-learn preprocessing API with methods: fit, transform, fit_transform and inverse_transform. Transformation is applied to each series before training the forecaster. ColumnTransformers are not allowed since they do not have inverse_transform method.

  • If single transformer: it is cloned and applied to all series.
  • If dict of transformers: a different transformer can be used for each series.
 None
transformer_exog transformer

An instance of a transformer (preprocessor) compatible with the scikit-learn preprocessing API. The transformation is applied to exog before training the forecaster. inverse_transform is not available when using ColumnTransformers.

 None
fit_kwargs dict

Additional arguments to be passed to the fit method of the estimator.

 `None`
forecaster_id (str, int)

Name used as an identifier of the forecaster.

 `None`
regressor estimator or pipeline compatible with the Keras API

Deprecated, alias for estimator.

 None

Attributes:

Name Type Description
estimator estimator or pipeline compatible with the Keras API

An instance of a estimator or pipeline compatible with the Keras API. An instance of this estimator is trained for each step. All of them are stored in self.estimators_.
levels (str, list)

Name of one or more time series to be predicted. This determine the series the forecaster will be handling. If None, all series used during training will be available for prediction.
layers_names list

Names of the layers in the Keras model used as estimator.
steps numpy ndarray

Future steps the forecaster will predict when using prediction methods.
max_step int

Maximum step the forecaster is able to predict. It is the maximum value included in steps.
lags numpy ndarray

Lags used as predictors.
max_lag int

Maximum lag included in lags.
window_size int

Size of the window needed to create the predictors.
transformer_series (object, dict)

An instance of a transformer (preprocessor) compatible with the scikit-learn preprocessing API with methods: fit, transform, fit_transform and inverse_transform. Transformation is applied to each series before training the forecaster. ColumnTransformers are not allowed since they do not have inverse_transform method.
transformer_series_ dict

Dictionary with the transformer for each series. It is created cloning the objects in transformer_series and is used internally to avoid overwriting.
transformer_exog transformer

An instance of a transformer (preprocessor) compatible with the scikit-learn preprocessing API. The transformation is applied to exog before training the forecaster. inverse_transform is not available when using ColumnTransformers.
last_window_ pandas DataFrame

This window represents the most recent data observed by the predictor during its training phase. It contains the values needed to predict the next step immediately after the training data. These values are stored in the original scale of the time series before undergoing any transformations or differentiation.
index_type_ type

Type of index of the input used in training.
index_freq_ str

Frequency of Index of the input used in training.
training_range_ pandas Index

First and last values of index of the data used during training.
series_names_in_ list

Names of the series used during training.
n_series_in int

Number of series used during training.
n_levels_out int

Number of levels (series) to be predicted by the forecaster.
exog_in_ bool

If the forecaster has been trained using exogenous variable/s.
exog_names_in_ list

Names of the exogenous variables used during training.
exog_type_in_ type

Type of exogenous variable/s used in training.
exog_dtypes_in_ dict

Type of each exogenous variable/s used in training before the transformation applied by transformer_exog. If transformer_exog is not used, it is equal to exog_dtypes_out_.
exog_dtypes_out_ dict

Type of each exogenous variable/s used in training after the transformation applied by transformer_exog. If transformer_exog is not used, it is equal to exog_dtypes_in_.
X_train_dim_names_ dict

Labels for the multi-dimensional arrays created internally for training.
y_train_dim_names_ dict

Labels for the multi-dimensional arrays created internally for training.
series_val pandas DataFrame

Values of the series used for validation during training.
exog_val pandas DataFrame

Values of the exogenous variables used for validation during training.
history dict

Dictionary with the history of the training of each step. It is created internally to avoid overwriting.
fit_kwargs dict

Additional arguments to be passed to the fit method of the estimator.
in_sample_residuals_ dict

Residuals of the model when predicting training data. Only stored up to 10_000 values per step in the form {step: residuals}. If transformer_series is not None, residuals are stored in the transformed scale.
out_sample_residuals_ dict

Residuals of the model when predicting non-training data. Only stored up to 10_000 values per step in the form {step: residuals}. Use set_out_sample_residuals() method to set values. If transformer_series is not None, residuals are stored in the transformed scale.
creation_date str

Date of creation.
is_fitted bool

Tag to identify if the estimator has been fitted (trained).
fit_date str

Date of last fit.
keras_backend_ str

Keras backend used to fit the forecaster. It can be 'tensorflow', 'torch' or 'jax'.
skforecast_version str

Version of skforecast library used to create the forecaster.
python_version str

Version of python used to create the forecaster.
forecaster_id (str, int)

Name used as an identifier of the forecaster.
__skforecast_tags__ dict

Tags associated with the forecaster.
_probabilistic_mode (str, bool)

Private attribute used to indicate whether the forecaster should perform some calculations during backtesting.
dropna_from_series Ignored

Not used, present here for API consistency by convention.
encoding Ignored

Not used, present here for API consistency by convention.
differentiation Ignored

Not used, present here for API consistency by convention.
differentiation_max Ignored

Not used, present here for API consistency by convention.
differentiator Ignored

Not used, present here for API consistency by convention.
differentiator_ Ignored

Not used, present here for API consistency by convention.

Methods:

Name Description
create_train_X_y

Create training matrices. The resulting multi-dimensional matrices contain
fit

Training Forecaster.
create_predict_X

Create the predictors needed to predict steps ahead.
predict

Predict n steps ahead
predict_interval

Predict n steps ahead and estimate prediction intervals using conformal

plot_history

Plots the training and validation loss curves from the given history object stored
set_params

Set new values to the parameters of the scikit-learn model stored in the
set_fit_kwargs

Set new values for the additional keyword arguments passed to the fit
set_lags

Not used, present here for API consistency by convention.
set_in_sample_residuals

Set in-sample residuals in case they were not calculated during the
set_out_sample_residuals

Set new values to the attribute out_sample_residuals_. Out of sample
Source code in skforecast/deep_learning/_forecaster_rnn.py 
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
def __init__(
    self,
    levels: str | list[str],
    lags: int | list[int] | np.ndarray[int] | range[int],
    estimator: object = None,
    transformer_series: object | dict[str, object] | None = MinMaxScaler(
        feature_range=(0, 1)
    ),
    transformer_exog: object | None = MinMaxScaler(feature_range=(0, 1)),
    fit_kwargs: dict[str, object] | None = {},
    forecaster_id: str | int | None = None,
    regressor: object = None
) -> None:

    self.estimator = deepcopy(initialize_estimator(estimator, regressor))
    self.levels = None
    self.transformer_series = transformer_series
    self.transformer_series_ = None
    self.transformer_exog = transformer_exog
    self.max_lag = None
    self.window_size = None
    self.last_window_ = None
    self.index_type_ = None
    self.index_freq_ = None
    self.training_range_ = None
    self.series_names_in_ = None
    self.exog_names_in_ = None
    self.exog_type_in_ = None
    self.exog_dtypes_in_ = None
    self.X_train_dim_names_ = None
    self.y_train_dim_names_ = None
    self.history_ = None 
    self.is_fitted = False
    self.creation_date = pd.Timestamp.today().strftime("%Y-%m-%d %H:%M:%S")
    self.fit_date = None
    self.keras_backend_ = None
    self.skforecast_version = __version__
    self.python_version = sys.version.split(" ")[0]
    self.forecaster_id = forecaster_id
    self._probabilistic_mode = "no_binned"

    self.weight_func = None  # Ignored in this forecaster
    self.source_code_weight_func = None  # Ignored in this forecaster
    self.dropna_from_series = False  # Ignored in this forecaster
    self.encoding = None  # Ignored in this forecaster
    self.in_sample_residuals_ = None  # Ignored in this forecaster
    self.in_sample_residuals_by_bin_ = None  # Ignored in this forecaster
    self.out_sample_residuals_ = None  # Ignored in this forecaster
    self.out_sample_residuals_by_bin_ = None  # Ignored in this forecaster
    self.differentiation = None  # Ignored in this forecaster
    self.differentiation_max = None  # Ignored in this forecaster
    self.differentiator = None  # Ignored in this forecaster
    self.differentiator_ = None  # Ignored in this forecaster

    layer_init = self.estimator.layers[0]
    layer_end = self.estimator.layers[-1]
    self.layers_names = [layer.name for layer in self.estimator.layers]

    self.lags, self.lags_names, self.max_lag = initialize_lags(
        type(self).__name__, lags
    )
    n_lags_estimator = layer_init.output.shape[1]
    if len(self.lags) != n_lags_estimator:
        raise ValueError(
            f"Number of lags ({len(self.lags)}) does not match the number of "
            f"lags expected by the estimator architecture ({n_lags_estimator})."
        )

    self.window_size = self.max_lag

    self.steps = np.arange(layer_end.output.shape[1]) + 1
    self.max_step = np.max(self.steps)

    if isinstance(levels, str):
        self.levels = [levels]
    elif isinstance(levels, list):
        self.levels = levels
    else:
        raise TypeError(
            f"`levels` argument must be a string or list. Got {type(levels)}."
        )

    self.n_series_in = self.estimator.get_layer('series_input').output.shape[-1]
    self.n_levels_out = self.estimator.get_layer('output_dense_td_layer').output.shape[-1]
    self.exog_in_ = True if "exog_input" in self.layers_names else False
    if self.exog_in_:
        self.n_exog_in = self.estimator.get_layer('exog_input').output.shape[-1]
    else:
        self.n_exog_in = None
        # NOTE: This is needed because the Reshape layer changes the output 
        # shape in _create_and_compile_model_no_exog
        self.n_levels_out = int(self.n_levels_out / self.max_step)

    if not len(self.levels) == self.n_levels_out:
        raise ValueError(
            f"Number of levels ({len(self.levels)}) does not match the number of "
            f"levels expected by the estimator architecture ({self.n_levels_out})."
        )

    self.series_val = None
    self.exog_val = None
    if "series_val" in fit_kwargs:
        if not isinstance(fit_kwargs["series_val"], pd.DataFrame):
            raise TypeError(
                f"`series_val` must be a pandas DataFrame. "
                f"Got {type(fit_kwargs['series_val'])}."
            )
        self.series_val = fit_kwargs.pop("series_val")

        if self.exog_in_:
            if "exog_val" not in fit_kwargs.keys():
                raise ValueError(
                    "If `series_val` is provided, `exog_val` must also be "
                    "provided using the `fit_kwargs` argument when the "
                    "estimator has exogenous variables."
                )
            else:
                if not isinstance(fit_kwargs["exog_val"], (pd.Series, pd.DataFrame)):
                    raise TypeError(
                        f"`exog_val` must be a pandas Series or DataFrame. "
                        f"Got {type(fit_kwargs['exog_val'])}."
                    )
                self.exog_val = input_to_frame(
                    data=fit_kwargs.pop("exog_val"), input_name='exog_val'
                )

    self.fit_kwargs = check_select_fit_kwargs(
        estimator=self.estimator, fit_kwargs=fit_kwargs
    )

    self.__skforecast_tags__ = {
        "library": "skforecast",
        "forecaster_name": "ForecasterRNN",
        "forecaster_task": "regression",
        "forecasting_scope": "global",  # single-series | global
        "forecasting_strategy": "deep_learning",  # recursive | direct | deep_learning
        "index_types_supported": ["pandas.RangeIndex", "pandas.DatetimeIndex"],
        "requires_index_frequency": True,

        "allowed_input_types_series": ["pandas.DataFrame"],
        "supports_exog": True,
        "allowed_input_types_exog": ["pandas.Series", "pandas.DataFrame"],
        "handles_missing_values_series": False, 
        "handles_missing_values_exog": False, 

        "supports_lags": True,
        "supports_window_features": False,
        "supports_transformer_series": True,
        "supports_transformer_exog": True,
        "supports_weight_func": False,
        "supports_series_weights": False,
        "supports_differentiation": False,

        "prediction_types": ["point", "interval"],
        "supports_probabilistic": True,
        "probabilistic_methods": ["conformal"],
        "handles_binned_residuals": False
    }

estimator instance-attribute 

estimator = deepcopy(
    initialize_estimator(estimator, regressor)
)

levels instance-attribute 

levels = None

transformer_series instance-attribute 

transformer_series = transformer_series

transformer_series_ instance-attribute 

transformer_series_ = None

transformer_exog instance-attribute 

transformer_exog = transformer_exog

max_lag instance-attribute 

max_lag = None

last_window_ instance-attribute 

last_window_ = None

index_type_ instance-attribute 

index_type_ = None

index_freq_ instance-attribute 

index_freq_ = None

training_range_ instance-attribute 

training_range_ = None

series_names_in_ instance-attribute 

series_names_in_ = None

exog_names_in_ instance-attribute 

exog_names_in_ = None

exog_type_in_ instance-attribute 

exog_type_in_ = None

exog_dtypes_in_ instance-attribute 

exog_dtypes_in_ = None

X_train_dim_names_ instance-attribute 

X_train_dim_names_ = None

y_train_dim_names_ instance-attribute 

y_train_dim_names_ = None

history_ instance-attribute 

history_ = None

is_fitted instance-attribute 

is_fitted = False

creation_date instance-attribute 

creation_date = strftime('%Y-%m-%d %H:%M:%S')

fit_date instance-attribute 

fit_date = None

keras_backend_ instance-attribute 

keras_backend_ = None

skforecast_version instance-attribute 

skforecast_version = __version__

python_version instance-attribute 

python_version = split(' ')[0]

forecaster_id instance-attribute 

forecaster_id = forecaster_id

_probabilistic_mode instance-attribute 

_probabilistic_mode = 'no_binned'

weight_func instance-attribute 

weight_func = None

source_code_weight_func instance-attribute 

source_code_weight_func = None

dropna_from_series instance-attribute 

dropna_from_series = False

encoding instance-attribute 

encoding = None

in_sample_residuals_ instance-attribute 

in_sample_residuals_ = None

in_sample_residuals_by_bin_ instance-attribute 

in_sample_residuals_by_bin_ = None

out_sample_residuals_ instance-attribute 

out_sample_residuals_ = None

out_sample_residuals_by_bin_ instance-attribute 

out_sample_residuals_by_bin_ = None

differentiation instance-attribute 

differentiation = None

differentiation_max instance-attribute 

differentiation_max = None

differentiator instance-attribute 

differentiator = None

differentiator_ instance-attribute 

differentiator_ = None

layers_names instance-attribute 

layers_names = [(name) for layer in (layers)]

window_size instance-attribute 

window_size = max_lag

steps instance-attribute 

steps = arange(shape[1]) + 1

max_step instance-attribute 

max_step = max(steps)

n_series_in instance-attribute 

n_series_in = shape[-1]

n_levels_out instance-attribute 

n_levels_out = shape[-1]

exog_in_ instance-attribute 

exog_in_ = True if 'exog_input' in layers_names else False

n_exog_in instance-attribute 

n_exog_in = shape[-1]

series_val instance-attribute 

series_val = None

exog_val instance-attribute 

exog_val = None

fit_kwargs instance-attribute 

fit_kwargs = check_select_fit_kwargs(
    estimator=estimator, fit_kwargs=fit_kwargs
)

_repr_html_ 

_repr_html_()

HTML representation of the object. The "General Information" section is expanded by default.

Source code in skforecast/deep_learning/_forecaster_rnn.py 
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
def _repr_html_(self) -> str:
    """
    HTML representation of the object.
    The "General Information" section is expanded by default.
    """

    params = str(self.estimator.get_config())
    compile_config = str(self.estimator.get_compile_config())

    (
        _,
        _,
        series_names_in_,
        exog_names_in_,
        transformer_series,
    ) = self._preprocess_repr(
            estimator          = None,
            series_names_in_   = self.series_names_in_,
            exog_names_in_     = self.exog_names_in_,
            transformer_series = self.transformer_series,
        )

    style, unique_id = get_style_repr_html(self.is_fitted)

    content = f"""
    <div class="container-{unique_id}">
        <p style="font-size: 1.5em; font-weight: bold; margin-block-start: 0.83em; margin-block-end: 0.83em;">{type(self).__name__}</p>
        <details open>
            <summary>General Information</summary>
            <ul>
                <li><strong>Estimator:</strong> {type(self.estimator).__name__}</li>
                <li><strong>Layers names:</strong> {self.layers_names}</li>
                <li><strong>Lags:</strong> {self.lags}</li>
                <li><strong>Window size:</strong> {self.window_size}</li>
                <li><strong>Maximum steps to predict:</strong> {self.steps}</li>
                <li><strong>Exogenous included:</strong> {self.exog_in_}</li>
                <li><strong>Creation date:</strong> {self.creation_date}</li>
                <li><strong>Last fit date:</strong> {self.fit_date}</li>
                <li><strong>Keras backend:</strong> {self.keras_backend_}</li>
                <li><strong>Skforecast version:</strong> {self.skforecast_version}</li>
                <li><strong>Python version:</strong> {self.python_version}</li>
                <li><strong>Forecaster id:</strong> {self.forecaster_id}</li>
            </ul>
        </details>
        <details>
            <summary>Exogenous Variables</summary>
            <ul>
                {exog_names_in_}
            </ul>
        </details>
        <details>
            <summary>Data Transformations</summary>
            <ul>
                <li><strong>Transformer for series:</strong> {transformer_series}</li>
                <li><strong>Transformer for exog:</strong> {self.transformer_exog}</li>
            </ul>
        </details>
        <details>
            <summary>Training Information</summary>
            <ul>
                <li><strong>Series names:</strong> {series_names_in_}</li>
                <li><strong>Target series (levels):</strong> {self.levels}</li>
                <li><strong>Training range:</strong> {self.training_range_.to_list() if self.is_fitted else 'Not fitted'}</li>
                <li><strong>Training index type:</strong> {str(self.index_type_).split('.')[-1][:-2] if self.is_fitted else 'Not fitted'}</li>
                <li><strong>Training index frequency:</strong> {self.index_freq_ if self.is_fitted else 'Not fitted'}</li>
            </ul>
        </details>
        <details>
            <summary>Estimator Parameters</summary>
            <ul>
                {params}
            </ul>
        </details>
        <details>
            <summary>Compile Parameters</summary>
            <ul>
                {compile_config}
            </ul>
        </details>
        <details>
            <summary>Fit Kwargs</summary>
            <ul>
                {self.fit_kwargs}
            </ul>
        </details>
        <p>
            <a href="https://skforecast.org/{__version__}/api/forecasterrnn.html">&#128712 <strong>API Reference</strong></a>
            &nbsp;&nbsp;
            <a href="https://skforecast.org/{__version__}/user_guides/forecasting-with-deep-learning-rnn-lstm.html">&#128462 <strong>User Guide</strong></a>
        </p>
    </div>
    """

    # Return the combined style and content
    return style + content

_create_lags 

_create_lags(y)

Transforms a 1d array into a 3d array (X) and a 3d array (y). Each row in X is associated with a value of y and it represents the lags that precede it.

Notice that, the returned matrix X_data, contains the lag 1 in the first column, the lag 2 in the second column and so on.

Parameters:

Name Type Description Default
y numpy ndarray

1d numpy ndarray Training time series.

 required

Returns:

Name Type Description
X_data numpy ndarray

3d numpy ndarray with the lagged values (predictors). Shape: (samples - max(lags), len(lags))
y_data numpy ndarray

3d numpy ndarray with the values of the time series related to each row of X_data for each step. Shape: (len(max_step), samples - max(lags))
Source code in skforecast/deep_learning/_forecaster_rnn.py 
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
def _create_lags(
    self, 
    y: np.ndarray
) -> tuple[np.ndarray, np.ndarray]:
    """
    Transforms a 1d array into a 3d array (X) and a 3d array (y). Each row
    in X is associated with a value of y and it represents the lags that
    precede it.

    Notice that, the returned matrix X_data, contains the lag 1 in the first
    column, the lag 2 in the second column and so on.

    Parameters
    ----------
    y : numpy ndarray
        1d numpy ndarray Training time series.

    Returns
    -------
    X_data : numpy ndarray
        3d numpy ndarray with the lagged values (predictors).
        Shape: (samples - max(lags), len(lags))
    y_data : numpy ndarray
        3d numpy ndarray with the values of the time series related to each
        row of `X_data` for each step.
        Shape: (len(max_step), samples - max(lags))

    """

    n_rows = len(y) - self.window_size - self.max_step + 1  # rows of y_data

    X_data = np.full(
        shape=(n_rows, (self.window_size)), fill_value=np.nan, order="F", dtype=float
    )
    for i, lag in enumerate(range(self.window_size - 1, -1, -1)):
        X_data[:, i] = y[self.window_size - lag - 1 : -(lag + self.max_step)]

    # Get lags index
    X_data = X_data[:, self.lags - 1]

    y_data = np.full(
        shape=(n_rows, self.max_step), fill_value=np.nan, order="F", dtype=float
    )
    for step in range(self.max_step):
        y_data[:, step] = y[self.window_size + step : self.window_size + step + n_rows]

    # Get steps index
    y_data = y_data[:, self.steps - 1]

    return X_data, y_data

_create_train_X_y 

_create_train_X_y(series, exog=None)

Create training matrices. The resulting multi-dimensional matrices contain the target variable and predictors needed to train the model.

Parameters:

Name Type Description Default
series pandas DataFrame

Training time series.

 required
exog pandas Series, pandas DataFrame

Exogenous variable/s included as predictor/s. Must have the same number of observations as series and their indexes must be aligned.

 None

Returns:

Name Type Description
X_train numpy ndarray

Training values (predictors) for each step. The resulting array has 3 dimensions: (n_observations, n_lags, n_series)
exog_train numpy ndarray

Value of exogenous variables aligned with X_train. (n_observations, n_exog)
y_train numpy ndarray

Values (target) of the time series related to each row of X_train. The resulting array has 3 dimensions: (n_observations, n_steps, n_levels)
dimension_names dict

Labels for the multi-dimensional arrays created internally for training.
exog_names_in_ list

Names of the exogenous variables included in the training matrices.
exog_dtypes_in_ dict

Type of each exogenous variable/s used in training before the transformation applied by transformer_exog. If transformer_exog is not used, it is equal to exog_dtypes_out_.
exog_dtypes_out_ dict

Type of each exogenous variable/s used in training after the transformation applied by transformer_exog. If transformer_exog is not used, it is equal to exog_dtypes_in_.
Source code in skforecast/deep_learning/_forecaster_rnn.py 
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
def _create_train_X_y(
    self,
    series: pd.DataFrame,
    exog: pd.Series | pd.DataFrame | None = None
) -> tuple[
    np.ndarray, 
    np.ndarray, 
    np.ndarray, 
    dict[int, list], 
    list[str], 
    dict[str, type], 
    dict[str, type]
]:
    """
    Create training matrices. The resulting multi-dimensional matrices contain
    the target variable and predictors needed to train the model.

    Parameters
    ----------
    series : pandas DataFrame
        Training time series.
    exog : pandas Series, pandas DataFrame, default None
        Exogenous variable/s included as predictor/s. Must have the same
        number of observations as `series` and their indexes must be aligned.

    Returns
    -------
    X_train : numpy ndarray
        Training values (predictors) for each step. The resulting array has
        3 dimensions: (n_observations, n_lags, n_series)
    exog_train: numpy ndarray
        Value of exogenous variables aligned with X_train. (n_observations, n_exog)
    y_train : numpy ndarray
        Values (target) of the time series related to each row of `X_train`.
        The resulting array has 3 dimensions: (n_observations, n_steps, n_levels)
    dimension_names : dict
        Labels for the multi-dimensional arrays created internally for training.
    exog_names_in_ : list
        Names of the exogenous variables included in the training matrices.
    exog_dtypes_in_ : dict
        Type of each exogenous variable/s used in training before the transformation
        applied by `transformer_exog`. If `transformer_exog` is not used, it
        is equal to `exog_dtypes_out_`.
    exog_dtypes_out_ : dict
        Type of each exogenous variable/s used in training after the transformation 
        applied by `transformer_exog`. If `transformer_exog` is not used, it 
        is equal to `exog_dtypes_in_`.

    """

    if not isinstance(series, pd.DataFrame):
        raise TypeError(f"`series` must be a pandas DataFrame. Got {type(series)}.")

    _, series_index = check_extract_values_and_index(
        data=series, data_label="`series`", return_values=False
    )
    series_names_in_ = list(series.columns)
    if not len(series_names_in_) == self.n_series_in:
        raise ValueError(
            f"Number of series in `series` ({len(series_names_in_)}) "
            f"does not match the number of series expected by the model "
            f"architecture ({self.n_series_in})."
        )

    if not set(self.levels).issubset(set(series_names_in_)):
        raise ValueError(
            f"`levels` defined when initializing the forecaster must be "
            f"included in `series` used for training. "
            f"{set(self.levels) - set(series_names_in_)} not found."
        )

    if len(series) < self.window_size + self.max_step:
        raise ValueError(
            f"Minimum length of `series` for training this forecaster is "
            f"{self.window_size + self.max_step}. Reduce the number of "
            f"predicted steps, {self.max_step}, or the maximum "
            f"lag, {self.max_lag}, if no more data is available.\n"
            f"    Length `series`: {len(series)}.\n"
            f"    Max step : {self.max_step}.\n"
            f"    Lags window size: {self.max_lag}."
        )

    if exog is None and self.exog_in_:
        raise ValueError(
            "The estimator architecture expects exogenous variables during "
            "training. Please provide the `exog` argument. If this is "
            "unexpected, check your estimator architecture or the "
            "initialization parameters of the forecaster."
        )
    if exog is not None and not self.exog_in_:
        raise ValueError(
            "Exogenous variables (`exog`) were provided, but the model "
            "architecture was not built to expect exogenous variables. Please "
            "remove the `exog` argument or rebuild the model to include "
            "exogenous inputs."
        )

    fit_transformer = False
    if not self.is_fitted:
        fit_transformer = True
        self.transformer_series_ = initialize_transformer_series(
                                       forecaster_name    = type(self).__name__,
                                       series_names_in_   = series_names_in_,
                                       transformer_series = self.transformer_series
                                   )

    # Step 1: Create lags for all columns
    X_train = []
    y_train = []

    # TODO: Add method argument to calculate lags and/or steps
    for serie in series_names_in_:
        x = series[serie]
        check_y(y=x)
        x = transform_series(
            series=x,
            transformer=self.transformer_series_[serie],
            fit=fit_transformer,
            inverse_transform=False,
        )
        X, _ = self._create_lags(x)
        X_train.append(X)

    for level in self.levels:
        y = series[level]
        check_y(y=y)
        y = transform_series(
            series=y,
            transformer=self.transformer_series_[level],
            fit=fit_transformer,
            inverse_transform=False,
        )

        _, y = self._create_lags(y)
        y_train.append(y)

    X_train = np.stack(X_train, axis=2)
    y_train = np.stack(y_train, axis=2)

    train_index = series_index[
        self.max_lag : (len(series_index) - self.max_step + 1)
    ]
    dimension_names = {
        "X_train": {
            0: train_index,
            1: self.lags_names[::-1],
            2: series_names_in_,
        },
        "y_train": {
            0: train_index,
            1: [f"step_{step}" for step in self.steps],
            2: self.levels,
        },
    }

    if exog is not None:

        check_exog(exog=exog, allow_nan=False)
        exog = input_to_frame(data=exog, input_name='exog')
        _, exog_index = check_extract_values_and_index(
            data=exog, data_label='`exog`', ignore_freq=True, return_values=False
        )

        if len(exog.columns) != self.n_exog_in:
            raise ValueError(
                f"Number of columns in `exog` ({len(exog.columns)}) "
                f"does not match the number of exogenous variables expected "
                f"by the model architecture ({self.n_exog_in})."
            )

        series_index_no_ws = series_index[self.window_size:]
        len_series = len(series)
        len_series_no_ws = len_series - self.window_size
        len_exog = len(exog)
        if not len_exog == len_series and not len_exog == len_series_no_ws:
            raise ValueError(
                f"Length of `exog` must be equal to the length of `series` (if "
                f"index is fully aligned) or length of `series` - `window_size` "
                f"(if `exog` starts after the first `window_size` values).\n"
                f"    `exog`                   : ({exog_index[0]} -- {exog_index[-1]})  (n={len_exog})\n"
                f"    `series`                 : ({series.index[0]} -- {series.index[-1]})  (n={len_series})\n"
                f"    `series` - `window_size` : ({series_index_no_ws[0]} -- {series_index_no_ws[-1]})  (n={len_series_no_ws})"
            )

        exog_names_in_ = exog.columns.to_list()
        if len(set(exog_names_in_) - set(series_names_in_)) != len(exog_names_in_):
            raise ValueError(
                f"`exog` cannot contain a column named the same as one of "
                f"the series (column names of series).\n"
                f"  `series` columns : {series_names_in_}.\n"
                f"  `exog`   columns : {exog_names_in_}."
            )

        exog_n_dim_in = len(exog_names_in_)
        exog_dtypes_in_ = get_exog_dtypes(exog=exog)
        exog = transform_dataframe(
            df=exog,
            transformer=self.transformer_exog,
            fit=fit_transformer,
            inverse_transform=False,
        )
        exog_n_dim_out = len(exog.columns)
        exog_dtypes_out_ = get_exog_dtypes(exog=exog)

        if exog_n_dim_in != exog_n_dim_out:
            raise ValueError(
                f"Number of columns in `exog` after transformation ({exog_n_dim_out}) "
                f"does not match the number of columns before transformation ({exog_n_dim_in}). "
                f"The ForecasterRnn does not support transformations that "
                f"change the number of columns in `exog`. Preprocess `exog` "
                f"before passing it to the `create_and_compile_model` function."
            )

        if len_exog == len_series:
            if not (exog_index == series_index).all():
                raise ValueError(
                    "When `exog` has the same length as `series`, the index "
                    "of `exog` must be aligned with the index of `series` "
                    "to ensure the correct alignment of values."
                )
        else:
            if not (exog_index == series_index_no_ws).all():
                raise ValueError(
                    "When `exog` doesn't contain the first `window_size` "
                    "observations, the index of `exog` must be aligned with "
                    "the index of `series` minus the first `window_size` "
                    "observations to ensure the correct alignment of values."
                )

        exog_train = []
        for _, exog_name in enumerate(exog.columns):
            _, exog_step = self._create_lags(exog[exog_name])
            exog_train.append(exog_step)

        exog_train = np.stack(exog_train, axis=2)

        dimension_names["exog_train"] = {
            0: train_index,
            1: [f"step_{step}" for step in self.steps],
            2: exog.columns.to_list(),
        }
    else:
        exog_train = None
        exog_names_in_ = None
        exog_dtypes_in_ = None
        exog_dtypes_out_ = None
        dimension_names["exog_train"] = {
            0: None,
            1: None,
            2: None
        }

    return (
        X_train, 
        exog_train, 
        y_train, 
        dimension_names,
        exog_names_in_,
        exog_dtypes_in_,
        exog_dtypes_out_
    )

create_train_X_y 

create_train_X_y(
    series, exog=None, suppress_warnings=False
)

Create training matrices. The resulting multi-dimensional matrices contain the target variable and predictors needed to train the model.

Parameters:

Name Type Description Default
series pandas DataFrame

Training time series.

 required
exog pandas Series, pandas DataFrame

Exogenous variable/s included as predictor/s. Must have the same number of observations as series and their indexes must be aligned.

 None
suppress_warnings bool

If True, skforecast warnings will be suppressed during the creation of the training matrices. See skforecast.exceptions.warn_skforecast_categories for more information.

 False

Returns:

Name Type Description
X_train numpy ndarray

Training values (predictors) for each step. The resulting array has 3 dimensions: (n_observations, n_lags, n_series)
exog_train numpy ndarray

Value of exogenous variables aligned with X_train. (n_observations, n_exog)
y_train numpy ndarray

Values (target) of the time series related to each row of X_train. The resulting array has 3 dimensions: (n_observations, n_steps, n_levels)
dimension_names dict

Labels for the multi-dimensional arrays created internally for training.
Source code in skforecast/deep_learning/_forecaster_rnn.py 
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
def create_train_X_y(
    self,
    series: pd.DataFrame,
    exog: pd.Series | pd.DataFrame | None = None,
    suppress_warnings: bool = False
) -> tuple[np.ndarray, np.ndarray, np.ndarray, dict[int, list]]:
    """
    Create training matrices. The resulting multi-dimensional matrices contain
    the target variable and predictors needed to train the model.

    Parameters
    ----------
    series : pandas DataFrame
        Training time series.
    exog : pandas Series, pandas DataFrame, default None
        Exogenous variable/s included as predictor/s. Must have the same
        number of observations as `series` and their indexes must be aligned.
    suppress_warnings : bool, default False
        If `True`, skforecast warnings will be suppressed during the creation
        of the training matrices. See skforecast.exceptions.warn_skforecast_categories 
        for more information.

    Returns
    -------
    X_train : numpy ndarray
        Training values (predictors) for each step. The resulting array has
        3 dimensions: (n_observations, n_lags, n_series)
    exog_train: numpy ndarray
        Value of exogenous variables aligned with X_train. (n_observations, n_exog)
    y_train : numpy ndarray
        Values (target) of the time series related to each row of `X_train`.
        The resulting array has 3 dimensions: (n_observations, n_steps, n_levels)
    dimension_names : dict
        Labels for the multi-dimensional arrays created internally for training.

    """

    set_skforecast_warnings(suppress_warnings, action='ignore')

    output = self._create_train_X_y(series=series, exog=exog)

    X_train = output[0]
    exog_train = output[1]
    y_train = output[2]
    dimension_names = output[3]

    set_skforecast_warnings(suppress_warnings, action='default')

    return X_train, exog_train, y_train, dimension_names

fit 

fit(
    series,
    exog=None,
    store_last_window=True,
    store_in_sample_residuals=False,
    random_state=123,
    suppress_warnings=False,
)

Training Forecaster.

Additional arguments to be passed to the fit method of the estimator can be added with the fit_kwargs argument when initializing the forecaster.

Parameters:

Name Type Description Default
series pandas DataFrame

Training time series.

 required
exog pandas Series, pandas DataFrame

Exogenous variable/s included as predictor/s. Must have the same number of observations as series and their indexes must be aligned so that series[i] is regressed on exog[i].

 None
store_last_window bool

Whether or not to store the last window (last_window_) of training data.

 True
store_in_sample_residuals bool

If True, in-sample residuals will be stored in the forecaster object after fitting (in_sample_residuals_ attribute).

 False
random_state int

Set a seed for the random generator so that the stored sample residuals are always deterministic.

 123
suppress_warnings bool

If True, skforecast warnings will be suppressed during the prediction process. See skforecast.exceptions.warn_skforecast_categories for more information.

 False

Returns:

Type Description
None
Source code in skforecast/deep_learning/_forecaster_rnn.py 
 912
 913
 914
 915
 916
 917
 918
 919
 920
 921
 922
 923
 924
 925
 926
 927
 928
 929
 930
 931
 932
 933
 934
 935
 936
 937
 938
 939
 940
 941
 942
 943
 944
 945
 946
 947
 948
 949
 950
 951
 952
 953
 954
 955
 956
 957
 958
 959
 960
 961
 962
 963
 964
 965
 966
 967
 968
 969
 970
 971
 972
 973
 974
 975
 976
 977
 978
 979
 980
 981
 982
 983
 984
 985
 986
 987
 988
 989
 990
 991
 992
 993
 994
 995
 996
 997
 998
 999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
1076
1077
1078
1079
1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
1096
1097
def fit(
    self,
    series: pd.DataFrame,
    exog: pd.Series | pd.DataFrame = None,
    store_last_window: bool = True,
    store_in_sample_residuals: bool = False,
    random_state: int = 123,
    suppress_warnings: bool = False
) -> None:
    """
    Training Forecaster.

    Additional arguments to be passed to the `fit` method of the estimator
    can be added with the `fit_kwargs` argument when initializing the forecaster.

    Parameters
    ----------
    series : pandas DataFrame
        Training time series.
    exog : pandas Series, pandas DataFrame, default None
        Exogenous variable/s included as predictor/s. Must have the same
        number of observations as `series` and their indexes must be aligned so
        that series[i] is regressed on exog[i].
    store_last_window : bool, default True
        Whether or not to store the last window (`last_window_`) of training data.
    store_in_sample_residuals : bool, default False
        If `True`, in-sample residuals will be stored in the forecaster object
        after fitting (`in_sample_residuals_` attribute).
    random_state : int, default 123
        Set a seed for the random generator so that the stored sample 
        residuals are always deterministic.
    suppress_warnings : bool, default False
        If `True`, skforecast warnings will be suppressed during the prediction
        process. See skforecast.exceptions.warn_skforecast_categories for more
        information.

    Returns
    -------
    None

    """

    set_skforecast_warnings(suppress_warnings, action="ignore")

    # Reset values in case the forecaster has already been fitted.
    self.last_window_ = None
    self.index_type_ = None
    self.index_freq_ = None
    self.training_range_ = None
    self.series_names_in_ = None
    self.exog_names_in_ = None
    self.exog_type_in_ = None
    self.exog_dtypes_in_ = None
    self.exog_dtypes_out_ = None
    self.X_train_dim_names_ = None
    self.y_train_dim_names_ = None
    self.exog_train_dim_names_ = None
    self.in_sample_residuals_ = None
    self.is_fitted = False
    self.fit_date = None
    self.keras_backend_ = keras.backend.backend()

    (
        X_train,
        exog_train,
        y_train,
        dimension_names,
        exog_names_in_,
        exog_dtypes_in_,
        exog_dtypes_out_,
    ) = self._create_train_X_y(series=series, exog=exog)

    # NOTE: Need here to avoid refitting the transformer_series_ with the 
    # validation data.
    self.is_fitted = True
    series_names_in_ = dimension_names["X_train"][2]

    if self.keras_backend_ == "torch":

        import torch
        device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
        print(f"Using '{self.keras_backend_}' backend with device: {device}")

        torch_device = torch.device(device)
        X_train = torch.tensor(X_train).to(torch_device)
        y_train = torch.tensor(y_train).to(torch_device)
        if exog_train is not None:
            exog_train = torch.tensor(exog_train).to(torch_device)

    if self.series_val is not None:
        series_val = self.series_val[series_names_in_]
        if exog is not None:
            exog_val = self.exog_val[exog_names_in_]
        else:
            exog_val = None

        X_val, exog_val, y_val, *_ = self._create_train_X_y(
            series=series_val, exog=exog_val
        )
        if self.keras_backend_ == "torch":
            X_val = torch.tensor(X_val).to(torch_device)
            y_val = torch.tensor(y_val).to(torch_device)
            if exog_val is not None:
                exog_val = torch.tensor(exog_val).to(torch_device)

        if self.exog_val is not None:
            history = self.estimator.fit(
                x=[X_train, exog_train],
                y=y_train,
                validation_data=([X_val, exog_val], y_val),
                **self.fit_kwargs,
            )
        else:
            history = self.estimator.fit(
                x=X_train,
                y=y_train,
                validation_data=(X_val, y_val),
                **self.fit_kwargs,
            )

    else:
        history = self.estimator.fit(
            x=X_train if exog_train is None else [X_train, exog_train],
            y=y_train,
            **self.fit_kwargs,
        )

    # TODO: Include binning in the forecaster
    self.in_sample_residuals_ = {}
    if store_in_sample_residuals:

        # NOTE: Convert to numpy array if using torch backend
        if self.keras_backend_ == "torch":
            y_train = y_train.detach().cpu().numpy()

        residuals = y_train - self.estimator.predict(
            x=X_train if exog_train is None else [X_train, exog_train], verbose=0
        )

        residuals = np.concatenate(
            [residuals[:, i, :] for i, step in enumerate(self.steps)]
        )

        rng = np.random.default_rng(seed=random_state)
        for i, level in enumerate(self.levels):
            residuals_level = residuals[:, i]
            if len(residuals_level) > 10_000:
                residuals_level = residuals_level[
                    rng.integers(low=0, high=len(residuals_level), size=10_000)
                ]
            self.in_sample_residuals_[level] = residuals_level
    else:
        for level in self.levels:
            self.in_sample_residuals_[level] = None

    self.series_names_in_ = series_names_in_
    self.X_train_series_names_in_ = series_names_in_
    self.X_train_dim_names_ = dimension_names["X_train"]
    self.y_train_dim_names_ = dimension_names["y_train"]
    self.history_ = history.history

    self.fit_date = pd.Timestamp.today().strftime("%Y-%m-%d %H:%M:%S")
    self.training_range_ = series.index[[0, -1]]
    self.index_type_ = type(series.index)
    if isinstance(series.index, pd.DatetimeIndex):
        self.index_freq_ = series.index.freq
    else:
        self.index_freq_ = series.index.step

    if exog is not None:
        # NOTE: self.exog_in_ is determined by the estimator architecture and
        # set during initialization.
        self.exog_names_in_ = exog_names_in_
        self.exog_type_in_ = type(exog)
        self.exog_dtypes_in_ = exog_dtypes_in_
        self.exog_dtypes_out_ = exog_dtypes_out_
        self.exog_train_dim_names_ = dimension_names["exog_train"]
        self.X_train_exog_names_out_ = dimension_names["exog_train"][2]
        self.X_train_features_names_out_ = dimension_names["X_train"][1] + dimension_names["exog_train"][2]
    else:
        self.X_train_features_names_out_ = dimension_names["X_train"][1]

    if store_last_window:
        self.last_window_ = series.iloc[-self.max_lag :, :].copy()

    set_skforecast_warnings(suppress_warnings, action="default")

_create_predict_inputs 

_create_predict_inputs(
    steps=None,
    levels=None,
    last_window=None,
    exog=None,
    predict_probabilistic=False,
    use_in_sample_residuals=True,
    check_inputs=True,
)

Create the inputs needed for the prediction process.

Parameters:

Name Type Description Default
steps (int, list)

Predict n steps. The value of steps must be less than or equal to the value of steps defined in the estimator architecture.

  • If int: Only steps within the range of 1 to int are predicted.
  • If list: List of ints. Only the steps contained in the list are predicted.
  • If None: As many steps are predicted as defined in the estimator architecture.
 None
levels (str, list)

Name(s) of the time series to be predicted. It must be included in levels, defined when initializing the forecaster. If None, all all series used during training will be available for prediction.

 None
last_window pandas Series, pandas DataFrame

Series values used to create the predictors (lags) needed to predict steps. If last_window = None, the values stored in self.last_window_ are used to calculate the initial predictors, and the predictions start right after training data.

 None
exog pandas Series, pandas DataFrame

Exogenous variable/s included as predictor/s.

 None
predict_probabilistic bool

If True, the necessary checks for probabilistic predictions will be performed.

 False
use_in_sample_residuals bool

If True, residuals from the training data are used as proxy of prediction error to create predictions. If False, out of sample residuals (calibration) are used. Out-of-sample residuals must be precomputed using Forecaster's set_out_sample_residuals() method.

 True
check_inputs bool

If True, the input is checked for possible warnings and errors with the check_predict_input function. This argument is created for internal use and is not recommended to be changed.

 True

Returns:

Name Type Description
X list

List of numpy arrays needed for prediction. The first element is the matrix of lags and the second element is the matrix of exogenous variables.
X_predict_dimension_names dict

Labels for the multi-dimensional arrays created internally for prediction.
steps list

Steps to predict.
levels list

Levels (series) to predict.
prediction_index pandas Index

Index of the predictions.
Source code in skforecast/deep_learning/_forecaster_rnn.py 
1099
1100
1101
1102
1103
1104
1105
1106
1107
1108
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
1121
1122
1123
1124
1125
1126
1127
1128
1129
1130
1131
1132
1133
1134
1135
1136
1137
1138
1139
1140
1141
1142
1143
1144
1145
1146
1147
1148
1149
1150
1151
1152
1153
1154
1155
1156
1157
1158
1159
1160
1161
1162
1163
1164
1165
1166
1167
1168
1169
1170
1171
1172
1173
1174
1175
1176
1177
1178
1179
1180
1181
1182
1183
1184
1185
1186
1187
1188
1189
1190
1191
1192
1193
1194
1195
1196
1197
1198
1199
1200
1201
1202
1203
1204
1205
1206
1207
1208
1209
1210
1211
1212
1213
1214
1215
1216
1217
1218
1219
1220
1221
1222
1223
1224
1225
1226
1227
1228
1229
1230
1231
1232
1233
1234
1235
1236
1237
1238
1239
1240
1241
1242
1243
1244
1245
1246
1247
1248
1249
1250
1251
1252
1253
1254
1255
1256
1257
1258
1259
1260
1261
1262
1263
1264
1265
1266
1267
1268
1269
1270
1271
1272
1273
1274
1275
1276
1277
1278
1279
1280
1281
1282
def _create_predict_inputs(
    self,
    steps: int | list[int] | None = None,
    levels: str | list[str] | None = None,
    last_window: pd.Series | pd.DataFrame | None = None,
    exog: pd.Series | pd.DataFrame | None = None,
    predict_probabilistic: bool = False,
    use_in_sample_residuals: bool = True,
    check_inputs: bool = True
) -> tuple[list[np.ndarray], dict[str, dict], list[int], list[str], pd.Index]:
    """
    Create the inputs needed for the prediction process.

    Parameters
    ----------
    steps : int, list, default None
        Predict n steps. The value of `steps` must be less than or equal to the 
        value of steps defined in the estimator architecture.

        - If `int`: Only steps within the range of 1 to int are predicted.
        - If `list`: List of ints. Only the steps contained in the list 
        are predicted.
        - If `None`: As many steps are predicted as defined in the estimator
        architecture.
    levels : str, list, default None
        Name(s) of the time series to be predicted. It must be included
        in `levels`, defined when initializing the forecaster. If `None`, all
        all series used during training will be available for prediction.
    last_window : pandas Series, pandas DataFrame, default None
        Series values used to create the predictors (lags) needed to 
        predict `steps`.
        If `last_window = None`, the values stored in `self.last_window_` are
        used to calculate the initial predictors, and the predictions start
        right after training data.
    exog : pandas Series, pandas DataFrame, default None
        Exogenous variable/s included as predictor/s.
    predict_probabilistic : bool, default False
        If `True`, the necessary checks for probabilistic predictions will be 
        performed.
    use_in_sample_residuals : bool, default True
        If `True`, residuals from the training data are used as proxy of
        prediction error to create predictions. 
        If `False`, out of sample residuals (calibration) are used. 
        Out-of-sample residuals must be precomputed using Forecaster's
        `set_out_sample_residuals()` method.
    check_inputs : bool, default True
        If `True`, the input is checked for possible warnings and errors 
        with the `check_predict_input` function. This argument is created 
        for internal use and is not recommended to be changed.

    Returns
    -------
    X : list
        List of numpy arrays needed for prediction. The first element
        is the matrix of lags and the second element is the matrix of
        exogenous variables.
    X_predict_dimension_names : dict
        Labels for the multi-dimensional arrays created internally for prediction.
    steps : list
        Steps to predict.
    levels : list
        Levels (series) to predict.
    prediction_index : pandas Index
        Index of the predictions.

    """

    levels, _ = prepare_levels_multiseries(
        X_train_series_names_in_=self.levels, levels=levels
    )

    steps = prepare_steps_direct(
                max_step = self.steps,
                steps    = steps
            )

    if last_window is None:
        last_window = self.last_window_

    if check_inputs:
        check_predict_input(
            forecaster_name=type(self).__name__,
            steps=steps,
            is_fitted=self.is_fitted,
            exog_in_=self.exog_in_,
            index_type_=self.index_type_,
            index_freq_=self.index_freq_,
            window_size=self.window_size,
            last_window=last_window,
            exog=exog,
            exog_names_in_=self.exog_names_in_,
            interval=None,
            max_step=self.max_step,
            levels=levels,
            levels_forecaster=self.levels,
            series_names_in_=self.series_names_in_,
        )

        if predict_probabilistic:
            check_residuals_input(
                forecaster_name              = type(self).__name__,
                use_in_sample_residuals      = use_in_sample_residuals,
                in_sample_residuals_         = self.in_sample_residuals_,
                out_sample_residuals_        = self.out_sample_residuals_,
                use_binned_residuals         = False,
                in_sample_residuals_by_bin_  = None,
                out_sample_residuals_by_bin_ = None,
                levels                       = self.levels
            )

    last_window = last_window.iloc[
        -self.window_size :, last_window.columns.get_indexer(self.series_names_in_)
    ].copy()

    last_window_values = last_window.to_numpy()
    last_window_matrix = np.full(
        shape=last_window.shape, fill_value=np.nan, order='F', dtype=float
    )
    for idx_series, series in enumerate(self.series_names_in_):
        last_window_series = last_window_values[:, idx_series]
        last_window_series = transform_numpy(
            array=last_window_series,
            transformer=self.transformer_series_[series],
            fit=False,
            inverse_transform=False,
        )
        last_window_matrix[:, idx_series] = last_window_series

    X = [np.reshape(last_window_matrix, (1, self.max_lag, last_window.shape[1]))]
    X_predict_dimension_names = {
        "X_autoreg": {
            0: "batch",
            1: self.lags_names[::-1],
            2: self.X_train_series_names_in_
        }
    }

    if exog is not None:

        exog = input_to_frame(data=exog, input_name='exog')
        exog = transform_dataframe(
            df=exog,
            transformer=self.transformer_exog,
            fit=False,
            inverse_transform=False,
        )

        exog_pred = exog.to_numpy()[:self.max_step]

        # NOTE: This is done to ensure that the exogenous variables
        # have the same number of rows as the maximum step to predict 
        # during backtesting when the last fold is incomplete 
        if len(exog_pred) < self.max_step:
            exog_pred = np.concatenate(
                [
                    exog_pred,
                    np.full(
                        shape=(self.max_step - len(exog_pred), exog_pred.shape[1]),
                        fill_value=0.,
                        dtype=float
                    )
                ],
                axis=0
            )

        exog_pred = np.expand_dims(exog_pred, axis=0)
        X.append(exog_pred)

        X_predict_dimension_names["exog_pred"] = {
            0: "batch",
            1: [f"step_{step}" for step in self.steps],
            2: self.X_train_exog_names_out_
        }

    prediction_index = expand_index(
                           index = last_window.index,
                           steps = max(steps)
                       )[np.array(steps) - 1]
    if isinstance(last_window.index, pd.DatetimeIndex) and np.array_equal(
        steps, np.arange(min(steps), max(steps) + 1)
    ):
        prediction_index.freq = last_window.index.freq

    return X, X_predict_dimension_names, steps, levels, prediction_index

create_predict_X 

create_predict_X(
    steps=None,
    levels=None,
    last_window=None,
    exog=None,
    suppress_warnings=False,
    check_inputs=True,
)

Create the predictors needed to predict steps ahead.

Parameters:

Name Type Description Default
steps (int, list)

Predict n steps. The value of steps must be less than or equal to the value of steps defined in the estimator architecture.

  • If int: Only steps within the range of 1 to int are predicted.
  • If list: List of ints. Only the steps contained in the list are predicted.
  • If None: As many steps are predicted as defined in the estimator architecture.
 None
levels (str, list)

Name(s) of the time series to be predicted. It must be included in levels, defined when initializing the forecaster. If None, all all series used during training will be available for prediction.

 None
last_window pandas DataFrame

Series values used to create the predictors (lags) needed to predict steps. If last_window = None, the values stored in self.last_window_ are used to calculate the initial predictors, and the predictions start right after training data.

 None
exog pandas Series, pandas DataFrame

Exogenous variable/s included as predictor/s.

 None
suppress_warnings bool

If True, skforecast warnings will be suppressed during the prediction process. See skforecast.exceptions.warn_skforecast_categories for more information.

 False
check_inputs bool

If True, the input is checked for possible warnings and errors with the check_predict_input function. This argument is created for internal use and is not recommended to be changed.

 True

Returns:

Name Type Description
X_predict pandas DataFrame

Pandas DataFrame with the predictors for each step.
exog_predict pandas DataFrame

Pandas DataFrame with the exogenous variables for each step.
Source code in skforecast/deep_learning/_forecaster_rnn.py 
1284
1285
1286
1287
1288
1289
1290
1291
1292
1293
1294
1295
1296
1297
1298
1299
1300
1301
1302
1303
1304
1305
1306
1307
1308
1309
1310
1311
1312
1313
1314
1315
1316
1317
1318
1319
1320
1321
1322
1323
1324
1325
1326
1327
1328
1329
1330
1331
1332
1333
1334
1335
1336
1337
1338
1339
1340
1341
1342
1343
1344
1345
1346
1347
1348
1349
1350
1351
1352
1353
1354
1355
1356
1357
1358
1359
1360
1361
1362
1363
1364
1365
1366
1367
1368
1369
1370
1371
1372
1373
1374
1375
1376
1377
1378
1379
1380
1381
1382
1383
1384
1385
def create_predict_X(
    self,
    steps: int | list[int] | None = None,
    levels: str | list[str] | None = None,
    last_window: pd.DataFrame | None = None,
    exog: pd.Series | pd.DataFrame | None = None,
    suppress_warnings: bool = False,
    check_inputs: bool = True
) -> tuple[pd.DataFrame, pd.DataFrame | None]:
    """
    Create the predictors needed to predict `steps` ahead.

    Parameters
    ----------
    steps : int, list, default None
        Predict n steps. The value of `steps` must be less than or equal to the 
        value of steps defined in the estimator architecture.

        - If `int`: Only steps within the range of 1 to int are predicted.
        - If `list`: List of ints. Only the steps contained in the list 
        are predicted.
        - If `None`: As many steps are predicted as defined in the estimator
        architecture.
    levels : str, list, default None
        Name(s) of the time series to be predicted. It must be included
        in `levels`, defined when initializing the forecaster. If `None`, all
        all series used during training will be available for prediction.
    last_window : pandas DataFrame, default None
        Series values used to create the predictors (lags) needed to 
        predict `steps`.
        If `last_window = None`, the values stored in `self.last_window_` are
        used to calculate the initial predictors, and the predictions start
        right after training data.
    exog : pandas Series, pandas DataFrame, default None
        Exogenous variable/s included as predictor/s.
    suppress_warnings : bool, default False
        If `True`, skforecast warnings will be suppressed during the prediction 
        process. See skforecast.exceptions.warn_skforecast_categories for more
        information.
    check_inputs : bool, default True
        If `True`, the input is checked for possible warnings and errors 
        with the `check_predict_input` function. This argument is created 
        for internal use and is not recommended to be changed.

    Returns
    -------
    X_predict : pandas DataFrame
        Pandas DataFrame with the predictors for each step.
    exog_predict : pandas DataFrame
        Pandas DataFrame with the exogenous variables for each step.

    """

    set_skforecast_warnings(suppress_warnings, action='ignore')

    (
        X,
        X_predict_dimension_names,
        *_
    ) = self._create_predict_inputs(
            steps        = steps,
            levels       = levels,
            last_window  = last_window,
            exog         = exog,
            check_inputs = check_inputs
        )

    X_predict = pd.DataFrame(
                    data    = X[0][0], 
                    columns = X_predict_dimension_names['X_autoreg'][2],
                    index   = X_predict_dimension_names['X_autoreg'][1] 
                )

    exog_predict = None
    if self.exog_in_:
        exog_predict = pd.DataFrame(
            data    = X[1][0], 
            columns = X_predict_dimension_names['exog_pred'][2],
            index   = X_predict_dimension_names['exog_pred'][1]
        )
        # NOTE: not needed in this forecaster
        # categorical_features = any(
        #     not pd.api.types.is_numeric_dtype(dtype) or pd.api.types.is_bool_dtype(dtype) 
        #     for dtype in set(self.exog_dtypes_out_.values())
        # )
        # if categorical_features:
        #     X_predict = X_predict.astype(self.exog_dtypes_out_)

    if self.transformer_series is not None:
        warnings.warn(
            "The output matrix is in the transformed scale due to the "
            "inclusion of transformations in the Forecaster. "
            "As a result, any predictions generated using this matrix will also "
            "be in the transformed scale. Please refer to the documentation "
            "for more details: "
            "https://skforecast.org/latest/user_guides/training-and-prediction-matrices.html",
            DataTransformationWarning
        )

    set_skforecast_warnings(suppress_warnings, action='default')

    return X_predict, exog_predict

predict 

predict(
    steps=None,
    levels=None,
    last_window=None,
    exog=None,
    suppress_warnings=False,
    check_inputs=True,
)

Predict n steps ahead

Parameters:

Name Type Description Default
steps (int, list)

Predict n steps. The value of steps must be less than or equal to the value of steps defined in the estimator architecture.

  • If int: Only steps within the range of 1 to int are predicted.
  • If list: List of ints. Only the steps contained in the list are predicted.
  • If None: As many steps are predicted as defined in the estimator architecture.
 None
levels (str, list)

Name(s) of the time series to be predicted. It must be included in levels, defined when initializing the forecaster. If None, all all series used during training will be available for prediction.

 None
last_window pandas DataFrame

Series values used to create the predictors (lags) needed in the first iteration of the prediction (t + 1). If last_window = None, the values stored in self.last_window_ are used to calculate the initial predictors, and the predictions start right after training data.

 `None`
exog pandas Series, pandas DataFrame

Exogenous variable/s included as predictor/s.

 None
suppress_warnings bool

If True, skforecast warnings will be suppressed during the fitting process. See skforecast.exceptions.warn_skforecast_categories for more information.

 `False`
check_inputs bool

If True, the input is checked for possible warnings and errors with the check_predict_input function. This argument is created for internal use and is not recommended to be changed.

 True

Returns:

Name Type Description
predictions pandas DataFrame

Predicted values.
Source code in skforecast/deep_learning/_forecaster_rnn.py 
1387
1388
1389
1390
1391
1392
1393
1394
1395
1396
1397
1398
1399
1400
1401
1402
1403
1404
1405
1406
1407
1408
1409
1410
1411
1412
1413
1414
1415
1416
1417
1418
1419
1420
1421
1422
1423
1424
1425
1426
1427
1428
1429
1430
1431
1432
1433
1434
1435
1436
1437
1438
1439
1440
1441
1442
1443
1444
1445
1446
1447
1448
1449
1450
1451
1452
1453
1454
1455
1456
1457
1458
1459
1460
1461
1462
1463
1464
1465
1466
1467
1468
1469
1470
1471
1472
1473
1474
1475
1476
1477
1478
1479
1480
1481
def predict(
    self,
    steps: int | list[int] | None = None,
    levels: str | list[str] | None = None,
    last_window: pd.DataFrame | None = None,
    exog: pd.Series | pd.DataFrame | None = None,
    suppress_warnings: bool = False,
    check_inputs: bool = True
) -> pd.DataFrame:
    """
    Predict n steps ahead

    Parameters
    ----------
    steps : int, list, default None
        Predict n steps. The value of `steps` must be less than or equal to the 
        value of steps defined in the estimator architecture.

        - If `int`: Only steps within the range of 1 to int are predicted.
        - If `list`: List of ints. Only the steps contained in the list 
        are predicted.
        - If `None`: As many steps are predicted as defined in the estimator
        architecture.
    levels : str, list, default None
        Name(s) of the time series to be predicted. It must be included
        in `levels`, defined when initializing the forecaster. If `None`, all
        all series used during training will be available for prediction.
    last_window : pandas DataFrame, default `None`
        Series values used to create the predictors (lags) needed in the
        first iteration of the prediction (t + 1).
        If `last_window = None`, the values stored in `self.last_window_` are
        used to calculate the initial predictors, and the predictions start
        right after training data.
    exog : pandas Series, pandas DataFrame, default None
        Exogenous variable/s included as predictor/s.
    suppress_warnings : bool, default `False`
        If `True`, skforecast warnings will be suppressed during the fitting
        process. See skforecast.exceptions.warn_skforecast_categories for more
        information.
    check_inputs : bool, default True
        If `True`, the input is checked for possible warnings and errors 
        with the `check_predict_input` function. This argument is created 
        for internal use and is not recommended to be changed.

    Returns
    -------
    predictions : pandas DataFrame
        Predicted values.

    """

    set_skforecast_warnings(suppress_warnings, action="ignore")

    (
        X,
        _,
        steps,
        levels,
        prediction_index
    ) = self._create_predict_inputs(
            steps        = steps,
            levels       = levels,
            last_window  = last_window,
            exog         = exog,
            check_inputs = check_inputs
        )

    predictions = self.estimator.predict(
        X[0] if not self.exog_in_ else X, verbose=0
    )
    predictions = np.reshape(
        predictions, (predictions.shape[1], predictions.shape[2])
    )[np.array(steps) - 1]

    for i, level in enumerate(self.levels):
        # NOTE: The inverse transformation is applied only if the level
        # is included in the levels to predict.
        if level in levels:
            predictions[:, i] = transform_numpy(
                array             = predictions[:, i],
                transformer       = self.transformer_series_[level],
                fit               = False,
                inverse_transform = True
            )

    n_steps, n_levels = predictions.shape
    predictions = pd.DataFrame(
        {"level": np.tile(self.levels, n_steps), "pred": predictions.ravel()},
        index = np.repeat(prediction_index, n_levels),
    )
    predictions = predictions[predictions['level'].isin(levels)]

    set_skforecast_warnings(suppress_warnings, action="default")

    return predictions

_predict_interval_conformal 

_predict_interval_conformal(
    steps=None,
    levels=None,
    last_window=None,
    exog=None,
    nominal_coverage=0.95,
    use_in_sample_residuals=True,
)

Generate prediction intervals using the conformal prediction split method [1]_.

Parameters:

Name Type Description Default
steps (int, list)

Predict n steps. The value of steps must be less than or equal to the value of steps defined in the estimator architecture.

  • If int: Only steps within the range of 1 to int are predicted.
  • If list: List of ints. Only the steps contained in the list are predicted.
  • If None: As many steps are predicted as defined in the estimator architecture.
 None
levels (str, list)

Name(s) of the time series to be predicted. It must be included in levels, defined when initializing the forecaster. If None, all all series used during training will be available for prediction.

 None
last_window pandas Series, pandas DataFrame

Series values used to create the predictors (lags) needed in the first iteration of the prediction (t + 1). If last_window = None, the values stored inself.last_window_ are used to calculate the initial predictors, and the predictions start right after training data.

 None
exog pandas Series, pandas DataFrame

Exogenous variable/s included as predictor/s.

 None
nominal_coverage float

Nominal coverage, also known as expected coverage, of the prediction intervals. Must be between 0 and 1.

 0.95
use_in_sample_residuals bool

If True, residuals from the training data are used as proxy of prediction error to create predictions. If False, out of sample residuals (calibration) are used. Out-of-sample residuals must be precomputed using Forecaster's set_out_sample_residuals() method.

 True

Returns:

Name Type Description
predictions pandas DataFrame

Values predicted by the forecaster and their estimated interval.

  • pred: predictions.
  • lower_bound: lower bound of the interval.
  • upper_bound: upper bound of the interval.
References

.. [1] MAPIE - Model Agnostic Prediction Interval Estimator. https://mapie.readthedocs.io/en/stable/theoretical_description_regression.html#the-split-method

Source code in skforecast/deep_learning/_forecaster_rnn.py 
1483
1484
1485
1486
1487
1488
1489
1490
1491
1492
1493
1494
1495
1496
1497
1498
1499
1500
1501
1502
1503
1504
1505
1506
1507
1508
1509
1510
1511
1512
1513
1514
1515
1516
1517
1518
1519
1520
1521
1522
1523
1524
1525
1526
1527
1528
1529
1530
1531
1532
1533
1534
1535
1536
1537
1538
1539
1540
1541
1542
1543
1544
1545
1546
1547
1548
1549
1550
1551
1552
1553
1554
1555
1556
1557
1558
1559
1560
1561
1562
1563
1564
1565
1566
1567
1568
1569
1570
1571
1572
1573
1574
1575
1576
1577
1578
1579
1580
1581
1582
1583
1584
1585
1586
1587
1588
1589
1590
1591
1592
1593
1594
1595
1596
1597
1598
1599
1600
1601
1602
1603
1604
1605
1606
1607
1608
1609
1610
1611
1612
1613
1614
1615
1616
def _predict_interval_conformal(
    self,
    steps: int | list[int] | None = None,
    levels: str | list[str] | None = None,
    last_window: pd.Series | pd.DataFrame | None = None,
    exog: pd.Series | pd.DataFrame | None = None,
    nominal_coverage: float = 0.95,
    use_in_sample_residuals: bool = True
) -> pd.DataFrame:
    """
    Generate prediction intervals using the conformal prediction 
    split method [1]_.

    Parameters
    ----------
    steps : int, list, default None
        Predict n steps. The value of `steps` must be less than or equal to the 
        value of steps defined in the estimator architecture.

        - If `int`: Only steps within the range of 1 to int are predicted.
        - If `list`: List of ints. Only the steps contained in the list 
        are predicted.
        - If `None`: As many steps are predicted as defined in the estimator
        architecture.
    levels : str, list, default None
        Name(s) of the time series to be predicted. It must be included
        in `levels`, defined when initializing the forecaster. If `None`, all
        all series used during training will be available for prediction.
    last_window : pandas Series, pandas DataFrame, default None
        Series values used to create the predictors (lags) needed in the 
        first iteration of the prediction (t + 1).
        If `last_window = None`, the values stored in` self.last_window_` are
        used to calculate the initial predictors, and the predictions start
        right after training data.
    exog : pandas Series, pandas DataFrame, default None
        Exogenous variable/s included as predictor/s.
    nominal_coverage : float, default 0.95
        Nominal coverage, also known as expected coverage, of the prediction
        intervals. Must be between 0 and 1.
    use_in_sample_residuals : bool, default True
        If `True`, residuals from the training data are used as proxy of
        prediction error to create predictions. 
        If `False`, out of sample residuals (calibration) are used. 
        Out-of-sample residuals must be precomputed using Forecaster's
        `set_out_sample_residuals()` method.

    Returns
    -------
    predictions : pandas DataFrame
        Values predicted by the forecaster and their estimated interval.

        - pred: predictions.
        - lower_bound: lower bound of the interval.
        - upper_bound: upper bound of the interval.

    References
    ----------
    .. [1] MAPIE - Model Agnostic Prediction Interval Estimator.
           https://mapie.readthedocs.io/en/stable/theoretical_description_regression.html#the-split-method

    """

    (
        X,
        _,
        steps,
        levels,
        prediction_index
    ) = self._create_predict_inputs(
            steps                   = steps,
            levels                  = levels,
            last_window             = last_window,
            exog                    = exog,
            predict_probabilistic   = True,
            use_in_sample_residuals = use_in_sample_residuals
        )

    if use_in_sample_residuals:
        residuals = self.in_sample_residuals_
    else:
        residuals = self.out_sample_residuals_

    predictions = self.estimator.predict(
        X[0] if not self.exog_in_ else X, verbose=0
    )
    predictions = np.reshape(
        predictions, (predictions.shape[1], predictions.shape[2])
    )[np.array(steps) - 1]

    n_steps = len(steps)
    n_levels = len(self.levels)
    correction_factor = np.full(
        shape=(n_steps, n_levels), fill_value=np.nan, order='C', dtype=float
    )
    for i, level in enumerate(self.levels):
        # NOTE: The correction factor is calculated only for the levels
        # included in the levels to predict.
        if level in levels:
            correction_factor[:, i] = np.quantile(
                np.abs(residuals[level]), nominal_coverage
            )
        else:
            correction_factor[:, i] = 0.

    lower_bound = predictions - correction_factor
    upper_bound = predictions + correction_factor

    # NOTE: Create a 3D array with shape (n_levels, intervals, steps)
    predictions = np.array([predictions, lower_bound, upper_bound]).swapaxes(0, 2)

    for i, level in enumerate(self.levels):
        # NOTE: The inverse transformation is applied only if the level
        # is included in the levels to predict.
        if level in levels:
            transformer_level = self.transformer_series_[level]
            if transformer_level is not None:
                predictions[i, :, :] = np.apply_along_axis(
                    func1d            = transform_numpy,
                    axis              = 0,
                    arr               = predictions[i, :, :],
                    transformer       = transformer_level,
                    fit               = False,
                    inverse_transform = True
                )

    predictions = pd.DataFrame(
                      data    = predictions.swapaxes(0, 1).reshape(-1, 3),
                      index   = np.repeat(prediction_index, len(self.levels)),
                      columns = ["pred", "lower_bound", "upper_bound"]
                  )
    predictions.insert(0, 'level', np.tile(self.levels, n_steps))
    predictions = predictions[predictions['level'].isin(levels)]

    return predictions

predict_interval 

predict_interval(
    steps=None,
    levels=None,
    last_window=None,
    exog=None,
    method="conformal",
    interval=[5, 95],
    use_in_sample_residuals=True,
    suppress_warnings=False,
    n_boot=None,
    use_binned_residuals=None,
    random_state=None,
)

Predict n steps ahead and estimate prediction intervals using conformal prediction method. Refer to the References section for additional details.

Parameters:

Name Type Description Default
steps (int, list)

Predict n steps. The value of steps must be less than or equal to the value of steps defined in the estimator architecture.

  • If int: Only steps within the range of 1 to int are predicted.
  • If list: List of ints. Only the steps contained in the list are predicted.
  • If None: As many steps are predicted as defined in the estimator architecture.
 None
levels (str, list)

Name(s) of the time series to be predicted. It must be included in levels, defined when initializing the forecaster. If None, all all series used during training will be available for prediction.

 None
last_window pandas DataFrame

Series values used to create the predictors (lags) needed in the first iteration of the prediction (t + 1). If last_window = None, the values stored in self.last_window_ are used to calculate the initial predictors, and the predictions start right after training data.

 None
exog pandas Series, pandas DataFrame, dict

Exogenous variable/s included as predictor/s.

 None
method str

Employs the conformal prediction split method for interval estimation [1]_.

 'conformal'
interval (float, list, tuple)

Confidence level of the prediction interval. Interpretation depends on the method used:

  • If float, represents the nominal (expected) coverage (between 0 and 1). For instance, interval=0.95 corresponds to [2.5, 97.5] percentiles.
  • If list or tuple, defines the exact percentiles to compute, which must be between 0 and 100 inclusive. For example, interval of 95% should be as interval = [2.5, 97.5].
  • When using method='conformal', the interval must be a float or a list/tuple defining a symmetric interval.
 [5, 95]
use_in_sample_residuals bool

If True, residuals from the training data are used as proxy of prediction error to create predictions. If False, out of sample residuals (calibration) are used. Out-of-sample residuals must be precomputed using Forecaster's set_out_sample_residuals() method.

 True
suppress_warnings bool

If True, skforecast warnings will be suppressed during the prediction process. See skforecast.exceptions.warn_skforecast_categories for more information.

 False
n_boot Ignored

Not used, present here for API consistency by convention.

 None
use_binned_residuals Ignored

Not used, present here for API consistency by convention.

 None
random_state Ignored

Not used, present here for API consistency by convention.

 None

Returns:

Name Type Description
predictions pandas DataFrame

Long-format DataFrame with the predictions and the lower and upper bounds of the estimated interval. The columns are level, pred, lower_bound, upper_bound.
References

.. [1] MAPIE - Model Agnostic Prediction Interval Estimator. https://mapie.readthedocs.io/en/stable/theoretical_description_regression.html#the-split-method

Source code in skforecast/deep_learning/_forecaster_rnn.py 
1618
1619
1620
1621
1622
1623
1624
1625
1626
1627
1628
1629
1630
1631
1632
1633
1634
1635
1636
1637
1638
1639
1640
1641
1642
1643
1644
1645
1646
1647
1648
1649
1650
1651
1652
1653
1654
1655
1656
1657
1658
1659
1660
1661
1662
1663
1664
1665
1666
1667
1668
1669
1670
1671
1672
1673
1674
1675
1676
1677
1678
1679
1680
1681
1682
1683
1684
1685
1686
1687
1688
1689
1690
1691
1692
1693
1694
1695
1696
1697
1698
1699
1700
1701
1702
1703
1704
1705
1706
1707
1708
1709
1710
1711
1712
1713
1714
1715
1716
1717
1718
1719
1720
1721
1722
1723
1724
1725
1726
1727
1728
1729
1730
def predict_interval(
    self,
    steps: int | list[int] | None = None,
    levels: str | list[str] | None = None,
    last_window: pd.DataFrame | None = None,
    exog: pd.Series | pd.DataFrame | None = None,
    method: str = 'conformal',
    interval: float | list[float] | tuple[float] = [5, 95],
    use_in_sample_residuals: bool = True,
    suppress_warnings: bool = False,
    n_boot: Any = None,
    use_binned_residuals: Any = None,
    random_state: Any = None,
) -> pd.DataFrame:
    """
    Predict n steps ahead and estimate prediction intervals using conformal 
    prediction method. Refer to the References section for additional details.

    Parameters
    ----------
    steps : int, list, default None
        Predict n steps. The value of `steps` must be less than or equal to the 
        value of steps defined in the estimator architecture.

        - If `int`: Only steps within the range of 1 to int are predicted.
        - If `list`: List of ints. Only the steps contained in the list 
        are predicted.
        - If `None`: As many steps are predicted as defined in the estimator
        architecture.
    levels : str, list, default None
        Name(s) of the time series to be predicted. It must be included
        in `levels`, defined when initializing the forecaster. If `None`, all
        all series used during training will be available for prediction.
    last_window : pandas DataFrame, default None
        Series values used to create the predictors (lags) needed in the 
        first iteration of the prediction (t + 1).
        If `last_window = None`, the values stored in `self.last_window_` are
        used to calculate the initial predictors, and the predictions start
        right after training data.
    exog : pandas Series, pandas DataFrame, dict, default None
        Exogenous variable/s included as predictor/s.
    method : str, default 'conformal'
        Employs the conformal prediction split method for interval estimation [1]_.
    interval : float, list, tuple, default [5, 95]
        Confidence level of the prediction interval. Interpretation depends 
        on the method used:

        - If `float`, represents the nominal (expected) coverage (between 0 
        and 1). For instance, `interval=0.95` corresponds to `[2.5, 97.5]` 
        percentiles.
        - If `list` or `tuple`, defines the exact percentiles to compute, which 
        must be between 0 and 100 inclusive. For example, interval 
        of 95% should be as `interval = [2.5, 97.5]`.
        - When using `method='conformal'`, the interval must be a float or 
        a list/tuple defining a symmetric interval.
    use_in_sample_residuals : bool, default True
        If `True`, residuals from the training data are used as proxy of
        prediction error to create predictions. 
        If `False`, out of sample residuals (calibration) are used. 
        Out-of-sample residuals must be precomputed using Forecaster's
        `set_out_sample_residuals()` method.
    suppress_warnings : bool, default False
        If `True`, skforecast warnings will be suppressed during the prediction 
        process. See skforecast.exceptions.warn_skforecast_categories for more
        information.
    n_boot : Ignored
        Not used, present here for API consistency by convention.
    use_binned_residuals : Ignored
        Not used, present here for API consistency by convention.
    random_state : Ignored
        Not used, present here for API consistency by convention.

    Returns
    -------
    predictions : pandas DataFrame
        Long-format DataFrame with the predictions and the lower and upper
        bounds of the estimated interval. The columns are `level`, `pred`,
        `lower_bound`, `upper_bound`.

    References
    ----------        
    .. [1] MAPIE - Model Agnostic Prediction Interval Estimator.
           https://mapie.readthedocs.io/en/stable/theoretical_description_regression.html#the-split-method

    """

    set_skforecast_warnings(suppress_warnings, action='ignore')

    if method == "conformal":

        if isinstance(interval, (list, tuple)):
            check_interval(interval=interval, ensure_symmetric_intervals=True)
            nominal_coverage = (interval[1] - interval[0]) / 100
        else:
            check_interval(alpha=interval, alpha_literal='interval')
            nominal_coverage = interval

        predictions = self._predict_interval_conformal(
                          steps                   = steps,
                          levels                  = levels,
                          last_window             = last_window,
                          exog                    = exog,
                          nominal_coverage        = nominal_coverage,
                          use_in_sample_residuals = use_in_sample_residuals
                      )
    else:
        raise ValueError(
            f"Invalid `method` '{method}'. Only 'conformal' is available."
        )

    set_skforecast_warnings(suppress_warnings, action='default')

    return predictions

plot_history 

plot_history(
    ax=None, exclude_first_iteration=False, **fig_kw
)

Plots the training and validation loss curves from the given history object stored in the ForecasterRnn.

Parameters:

Name Type Description Default
ax Axes

Pre-existing ax for the plot. Otherwise, call matplotlib.pyplot.subplots() internally.

 `None`
exclude_first_iteration bool

Whether to exclude the first epoch from the plot.

 `False`
fig_kw dict

Other keyword arguments are passed to matplotlib.pyplot.subplots().

 {}

Returns:

Name Type Description
fig Figure

Matplotlib Figure.
Source code in skforecast/deep_learning/_forecaster_rnn.py 
1732
1733
1734
1735
1736
1737
1738
1739
1740
1741
1742
1743
1744
1745
1746
1747
1748
1749
1750
1751
1752
1753
1754
1755
1756
1757
1758
1759
1760
1761
1762
1763
1764
1765
1766
1767
1768
1769
1770
1771
1772
1773
1774
1775
1776
1777
1778
1779
1780
1781
1782
1783
1784
1785
1786
1787
1788
1789
1790
1791
1792
1793
1794
1795
1796
1797
1798
1799
1800
1801
1802
1803
1804
1805
1806
1807
1808
def plot_history(
    self,
    ax: matplotlib.axes.Axes = None,
    exclude_first_iteration: bool = False,
    **fig_kw,
) -> matplotlib.figure.Figure:
    """
    Plots the training and validation loss curves from the given history object stored
    in the ForecasterRnn.

    Parameters
    ----------
    ax : matplotlib.axes.Axes, default `None`
        Pre-existing ax for the plot. Otherwise, call matplotlib.pyplot.subplots()
        internally.
    exclude_first_iteration : bool, default `False`
        Whether to exclude the first epoch from the plot.
    fig_kw : dict
        Other keyword arguments are passed to matplotlib.pyplot.subplots().

    Returns
    -------
    fig: matplotlib.figure.Figure
        Matplotlib Figure.

    """

    if ax is None:
        fig, ax = plt.subplots(1, 1, **fig_kw)
    else:
        fig = ax.get_figure()

    # Setting up the plot style
    if self.history_ is None:
        raise ValueError("ForecasterRnn has not been fitted yet.")

    # Determine the range of epochs to plot, excluding the first one if specified
    epoch_range = range(1, len(self.history_["loss"]) + 1)
    if exclude_first_iteration:
        epoch_range = range(2, len(self.history_["loss"]) + 1)

    # Plotting training loss
    ax.plot(
        epoch_range,
        self.history_["loss"][
            exclude_first_iteration:
        ],  # Skip first element if specified
        color="b",
        label="Training Loss",
    )

    # Plotting validation loss
    if "val_loss" in self.history_:
        ax.plot(
            epoch_range,
            self.history_["val_loss"][
                exclude_first_iteration:
            ],  # Skip first element if specified
            color="r",
            label="Validation Loss",
        )

    # Labeling the axes and adding a title
    ax.set_xlabel("Epochs")
    ax.set_ylabel("Loss")
    ax.set_title("Training and Validation Loss")

    # Adding a legend
    ax.legend()

    # Displaying grid for better readability
    ax.grid(True, linestyle="--", alpha=0.7)

    # Setting x-axis ticks to integers only
    ax.set_xticks(epoch_range)

    return fig

set_params 

set_params(params)

Set new values to the parameters of the scikit-learn model stored in the forecaster. It is important to note that all models share the same configuration of parameters and hyperparameters.

Parameters:

Name Type Description Default
params dict

Parameters values.

 required

Returns:

Type Description
None
Source code in skforecast/deep_learning/_forecaster_rnn.py 
1811
1812
1813
1814
1815
1816
1817
1818
1819
1820
1821
1822
1823
1824
1825
1826
1827
1828
1829
1830
def set_params(self, params: dict) -> None:  
    """
    Set new values to the parameters of the scikit-learn model stored in the
    forecaster. It is important to note that all models share the same
    configuration of parameters and hyperparameters.

    Parameters
    ----------
    params : dict
        Parameters values.

    Returns
    -------
    None

    """

    self.estimator = clone(self.estimator)
    self.estimator.reset_states()
    self.estimator.compile(**params)

set_fit_kwargs 

set_fit_kwargs(fit_kwargs)

Set new values for the additional keyword arguments passed to the fit method of the estimator.

Parameters:

Name Type Description Default
fit_kwargs dict

Dict of the form {"argument": new_value}.

 required

Returns:

Type Description
None
Source code in skforecast/deep_learning/_forecaster_rnn.py 
1832
1833
1834
1835
1836
1837
1838
1839
1840
1841
1842
1843
1844
1845
1846
1847
1848
1849
1850
1851
1852
1853
1854
1855
1856
1857
1858
1859
1860
1861
1862
1863
1864
1865
1866
1867
1868
1869
1870
1871
1872
1873
1874
1875
1876
1877
def set_fit_kwargs(self, fit_kwargs: dict) -> None:
    """
    Set new values for the additional keyword arguments passed to the `fit`
    method of the estimator.

    Parameters
    ----------
    fit_kwargs : dict
        Dict of the form {"argument": new_value}.

    Returns
    -------
    None

    """

    self.series_val = None
    self.exog_val = None
    if "series_val" in fit_kwargs:
        if not isinstance(fit_kwargs["series_val"], pd.DataFrame):
            raise TypeError(
                f"`series_val` must be a pandas DataFrame. "
                f"Got {type(fit_kwargs['series_val'])}."
            )
        self.series_val = fit_kwargs.pop("series_val")

        if self.exog_in_:
            if "exog_val" not in fit_kwargs.keys():
                raise ValueError(
                    "If `series_val` is provided, `exog_val` must also be "
                    "provided using the `fit_kwargs` argument when the "
                    "estimator has exogenous variables."
                )
            else:
                if not isinstance(fit_kwargs["exog_val"], (pd.Series, pd.DataFrame)):
                    raise TypeError(
                        f"`exog_val` must be a pandas Series or DataFrame. "
                        f"Got {type(fit_kwargs['exog_val'])}."
                    )
                self.exog_val = input_to_frame(
                    data=fit_kwargs.pop("exog_val"), input_name='exog_val'
                )

    self.fit_kwargs = check_select_fit_kwargs(
        self.estimator, fit_kwargs=fit_kwargs
    )

set_lags 

set_lags(lags)

Not used, present here for API consistency by convention.

Returns:

Type Description
None
Source code in skforecast/deep_learning/_forecaster_rnn.py 
1879
1880
1881
1882
1883
1884
1885
1886
1887
1888
1889
def set_lags(self, lags: Any) -> None:  # pragma: no cover
    """
    Not used, present here for API consistency by convention.

    Returns
    -------
    None

    """

    pass

set_in_sample_residuals 

set_in_sample_residuals(
    series,
    exog=None,
    random_state=123,
    suppress_warnings=False,
)

Set in-sample residuals in case they were not calculated during the training process.

In-sample residuals are calculated as the difference between the true values and the predictions made by the forecaster using the training data. The following internal attributes are updated:

  • in_sample_residuals_: Dictionary containing a numpy ndarray with the residuals for each series in the form {series: residuals}.

A total of 10_000 residuals are stored in the attribute in_sample_residuals_. If the number of residuals is greater than 10_000, a random sample of 10_000 residuals is stored. The number of residuals stored per bin is limited to 10_000 // self.binner.n_bins_.

Parameters:

Name Type Description Default
series pandas DataFrame

Training time series.

 required
exog pandas Series, pandas DataFrame

Exogenous variable/s included as predictor/s.

 None
random_state int

Sets a seed to the random sampling for reproducible output.

 123
suppress_warnings bool

If True, skforecast warnings will be suppressed during the sampling process. See skforecast.exceptions.warn_skforecast_categories for more information.

 False

Returns:

Type Description
None
Source code in skforecast/deep_learning/_forecaster_rnn.py 
1891
1892
1893
1894
1895
1896
1897
1898
1899
1900
1901
1902
1903
1904
1905
1906
1907
1908
1909
1910
1911
1912
1913
1914
1915
1916
1917
1918
1919
1920
1921
1922
1923
1924
1925
1926
1927
1928
1929
1930
1931
1932
1933
1934
1935
1936
1937
1938
1939
1940
1941
1942
1943
1944
1945
1946
1947
1948
1949
1950
1951
1952
1953
1954
1955
1956
1957
1958
1959
1960
1961
1962
1963
1964
1965
1966
1967
1968
1969
1970
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
def set_in_sample_residuals(
    self,
    series: pd.DataFrame,
    exog: pd.Series | pd.DataFrame = None,
    random_state: int = 123,
    suppress_warnings: bool = False
) -> None:
    """
    Set in-sample residuals in case they were not calculated during the
    training process. 

    In-sample residuals are calculated as the difference between the true 
    values and the predictions made by the forecaster using the training 
    data. The following internal attributes are updated:

    + `in_sample_residuals_`: Dictionary containing a numpy ndarray with the
    residuals for each series in the form `{series: residuals}`.

    A total of 10_000 residuals are stored in the attribute `in_sample_residuals_`.
    If the number of residuals is greater than 10_000, a random sample of
    10_000 residuals is stored. The number of residuals stored per bin is
    limited to `10_000 // self.binner.n_bins_`.

    Parameters
    ----------
    series : pandas DataFrame
        Training time series.
    exog : pandas Series, pandas DataFrame, default None
        Exogenous variable/s included as predictor/s.
    random_state : int, default 123
        Sets a seed to the random sampling for reproducible output.
    suppress_warnings : bool, default False
        If `True`, skforecast warnings will be suppressed during the sampling 
        process. See skforecast.exceptions.warn_skforecast_categories for more
        information.

    Returns
    -------
    None

    """

    set_skforecast_warnings(suppress_warnings, action='ignore')

    if not self.is_fitted:
        raise NotFittedError(
            "This forecaster is not fitted yet. Call `fit` with appropriate "
            "arguments before using `set_in_sample_residuals()`."
        )

    if not isinstance(series, pd.DataFrame):
        raise TypeError(
            f"`series` must be a pandas DataFrame. Got {type(series)}."
        )

    series_index_range = check_extract_values_and_index(
        data=series, data_label='`series`', return_values=False
    )[1][[0, -1]]
    if not series_index_range.equals(self.training_range_):
        raise IndexError(
            f"The index range of `series` does not match the range "
            f"used during training. Please ensure the index is aligned "
            f"with the training data.\n"
            f"    Expected : {self.training_range_}\n"
            f"    Received : {series_index_range}"
        )

    (
        X_train,
        exog_train,
        y_train,
        dimension_names,
        *_
    ) = self._create_train_X_y(series=series, exog=exog)

    if exog is not None:
        X_train_features_names_out_ = dimension_names["X_train"][1] + dimension_names["exog_train"][2]
    else:
        X_train_features_names_out_ = dimension_names["X_train"][1]

    if not X_train_features_names_out_ == self.X_train_features_names_out_:
        raise ValueError(
            f"Feature mismatch detected after matrix creation. The features "
            f"generated from the provided data do not match those used during "
            f"the training process. To correctly set in-sample residuals, "
            f"ensure that the same data and preprocessing steps are applied.\n"
            f"    Expected output : {self.X_train_features_names_out_}\n"
            f"    Current output  : {X_train_features_names_out_}"
        )

    # TODO: Include binning in the forecaster
    self.in_sample_residuals_ = {}
    residuals = y_train - self.estimator.predict(
        x=X_train if exog_train is None else [X_train, exog_train], verbose=0
    )
    residuals = np.concatenate(
        [residuals[:, i, :] for i, step in enumerate(self.steps)]
    )

    rng = np.random.default_rng(seed=random_state)
    for i, level in enumerate(self.levels):
        residuals_level = residuals[:, i]
        if len(residuals_level) > 10_000:
            residuals_level = residuals_level[
                rng.integers(low=0, high=len(residuals_level), size=10_000)
            ]
        self.in_sample_residuals_[level] = residuals_level

    set_skforecast_warnings(suppress_warnings, action='default')

set_out_sample_residuals 

set_out_sample_residuals(
    y_true, y_pred, append=False, random_state=123
)

Set new values to the attribute out_sample_residuals_. Out of sample residuals are meant to be calculated using observations that did not participate in the training process. y_true and y_pred are expected to be in the original scale of the time series. Residuals are calculated as y_true - y_pred, after applying the necessary transformations and differentiations if the forecaster includes them (self.transformer_series and self.differentiation).

A total of 10_000 residuals are stored in the attribute out_sample_residuals_. If the number of residuals is greater than 10_000, a random sample of 10_000 residuals is stored.

Parameters:

Name Type Description Default
y_true dict

Dictionary of numpy ndarrays or pandas Series with the true values of the time series for each series in the form {series: y_true}.

 required
y_pred dict

Dictionary of numpy ndarrays or pandas Series with the predicted values of the time series for each series in the form {series: y_pred}.

 required
append bool

If True, new residuals are added to the once already stored in the attribute out_sample_residuals_. If after appending the new residuals, the limit of 10_000 samples is exceeded, a random sample of 10_000 is kept.

 False
random_state int

Sets a seed to the random sampling for reproducible output.

 123

Returns:

Type Description
None
Source code in skforecast/deep_learning/_forecaster_rnn.py 
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
2021
2022
2023
2024
2025
2026
2027
2028
2029
2030
2031
2032
2033
2034
2035
2036
2037
2038
2039
2040
2041
2042
2043
2044
2045
2046
2047
2048
2049
2050
2051
2052
2053
2054
2055
2056
2057
2058
2059
2060
2061
2062
2063
2064
2065
2066
2067
2068
2069
2070
2071
2072
2073
2074
2075
2076
2077
2078
2079
2080
2081
2082
2083
2084
2085
2086
2087
2088
2089
2090
2091
2092
2093
2094
2095
2096
2097
2098
2099
2100
2101
2102
2103
2104
2105
2106
2107
2108
2109
2110
2111
2112
2113
2114
2115
2116
2117
2118
2119
2120
2121
2122
2123
2124
2125
2126
2127
2128
2129
2130
2131
2132
2133
2134
2135
2136
2137
2138
2139
2140
2141
2142
2143
2144
2145
2146
2147
2148
def set_out_sample_residuals(
    self,
    y_true: dict[str, np.ndarray | pd.Series],
    y_pred: dict[str, np.ndarray | pd.Series],
    append: bool = False,
    random_state: int = 123
) -> None:
    """
    Set new values to the attribute `out_sample_residuals_`. Out of sample
    residuals are meant to be calculated using observations that did not
    participate in the training process. `y_true` and `y_pred` are expected
    to be in the original scale of the time series. Residuals are calculated
    as `y_true` - `y_pred`, after applying the necessary transformations and
    differentiations if the forecaster includes them (`self.transformer_series`
    and `self.differentiation`).

    A total of 10_000 residuals are stored in the attribute `out_sample_residuals_`.
    If the number of residuals is greater than 10_000, a random sample of
    10_000 residuals is stored.

    Parameters
    ----------
    y_true : dict
        Dictionary of numpy ndarrays or pandas Series with the true values of
        the time series for each series in the form {series: y_true}.
    y_pred : dict
        Dictionary of numpy ndarrays or pandas Series with the predicted values
        of the time series for each series in the form {series: y_pred}.
    append : bool, default False
        If `True`, new residuals are added to the once already stored in the
        attribute `out_sample_residuals_`. If after appending the new residuals,
        the limit of 10_000 samples is exceeded, a random sample of 10_000 is
        kept.
    random_state : int, default 123
        Sets a seed to the random sampling for reproducible output.

    Returns
    -------
    None

    """

    if not self.is_fitted:
        raise NotFittedError(
            "This forecaster is not fitted yet. Call `fit` with appropriate "
            "arguments before using `set_out_sample_residuals()`."
        )

    if not isinstance(y_true, dict):
        raise TypeError(
            f"`y_true` must be a dictionary of numpy ndarrays or pandas Series. "
            f"Got {type(y_true)}."
        )

    if not isinstance(y_pred, dict):
        raise TypeError(
            f"`y_pred` must be a dictionary of numpy ndarrays or pandas Series. "
            f"Got {type(y_pred)}."
        )

    if not set(y_true.keys()) == set(y_pred.keys()):
        raise ValueError(
            f"`y_true` and `y_pred` must have the same keys. "
            f"Got {set(y_true.keys())} and {set(y_pred.keys())}."
        )

    for k in y_true.keys():
        if not isinstance(y_true[k], (np.ndarray, pd.Series)):
            raise TypeError(
                f"Values of `y_true` must be numpy ndarrays or pandas Series. "
                f"Got {type(y_true[k])} for series {k}."
            )
        if not isinstance(y_pred[k], (np.ndarray, pd.Series)):
            raise TypeError(
                f"Values of `y_pred` must be numpy ndarrays or pandas Series. "
                f"Got {type(y_pred[k])} for series {k}."
            )
        if len(y_true[k]) != len(y_pred[k]):
            raise ValueError(
                f"`y_true` and `y_pred` must have the same length. "
                f"Got {len(y_true[k])} and {len(y_pred[k])} for series {k}."
            )
        if isinstance(y_true[k], pd.Series) and isinstance(y_pred[k], pd.Series):
            if not y_true[k].index.equals(y_pred[k].index):
                raise ValueError(
                    f"When containing pandas Series, elements in `y_true` and "
                    f"`y_pred` must have the same index. Error in series {k}."
                )

    series_to_update = set(y_pred.keys()).intersection(set(self.levels))
    if not series_to_update:
        raise ValueError(
            f"Provided keys in `y_pred` and `y_true` do not match any of the "
            f"target time series in the forecaster, {self.levels}. Residuals "
            f"cannot be updated."
        )

    if self.out_sample_residuals_ is None:
        self.out_sample_residuals_ = {level: None for level in self.levels}

    rng = np.random.default_rng(seed=random_state)
    for level in series_to_update:

        y_true_level = deepcopy(y_true[level])
        y_pred_level = deepcopy(y_pred[level])
        if not isinstance(y_true_level, np.ndarray):
            y_true_level = y_true_level.to_numpy()
        if not isinstance(y_pred_level, np.ndarray):
            y_pred_level = y_pred_level.to_numpy()

        if self.transformer_series:
            y_true_level = transform_numpy(
                               array             = y_true_level,
                               transformer       = self.transformer_series_[level],
                               fit               = False,
                               inverse_transform = False
                           )
            y_pred_level = transform_numpy(
                               array             = y_pred_level,
                               transformer       = self.transformer_series_[level],
                               fit               = False,
                               inverse_transform = False
                           )

        data = pd.DataFrame(
            {'prediction': y_pred_level, 'residuals': y_true_level - y_pred_level}
        ).dropna()
        residuals = data['residuals'].to_numpy()

        out_sample_residuals = self.out_sample_residuals_.get(level, np.array([]))
        out_sample_residuals = (
            np.array([]) 
            if out_sample_residuals is None
            else out_sample_residuals
        )
        if append:
            out_sample_residuals = np.concatenate([out_sample_residuals, residuals])
        else:
            out_sample_residuals = residuals

        if len(out_sample_residuals) > 10_000:
            out_sample_residuals = rng.choice(
                a       = out_sample_residuals, 
                size    = 10_000, 
                replace = False
            )

        self.out_sample_residuals_[level] = out_sample_residuals