NBeatsForecaster

class NBeatsForecaster(window, horizon=1, backcast_loss_weight=0.5, batch_size=32, n_epochs=100, verbose=0, optimizer=None, metrics='mae', loss='mse', callbacks=None, random_state=None, axis=0, last_file_name='last_model', save_best_model=False, save_last_model=False, save_init_model=False, best_file_name='best_model', init_file_name='init_model', file_path='./', stacks=None, num_blocks_per_stack=2, units=30, num_trend_coefficients=3, num_seasonal_coefficients=5, num_generic_coefficients=7, share_weights=True, share_coefficients=True)[source]

Bases: BaseDeepForecaster, SeriesToSeriesForecastingMixin

N-BEATS deep learning forecaster.

Wraps NBeatsNetwork into the aeon forecasting API. Each block produces a backcast (input reconstruction) and a forecast (future prediction). Both heads are trained simultaneously; only the forecast is returned at inference time. Based on [1].

Parameters:
windowint

Lookback window size.

horizonint, default=1

Number of steps to forecast. Fixed at construction time.

backcast_loss_weightfloat, default=0.5

Loss weight for the backcast head relative to the forecast head (1.0).

batch_sizeint, default=32
n_epochsint, default=100
verboseint, default=0
optimizertf.keras.optimizers.Optimizer or None, default=None

Defaults to Adam.

metricsstr or list, default=”mae”

Metrics tracked on the forecast head during training.

lossstr or tf.keras.losses.Loss, default=”mse”

Applied to both forecast and backcast outputs.

callbackslist or None, default=None
random_stateint or None, default=None
axisint, default=0
last_file_namestr, default=”last_model”
save_best_modelbool, default=False
save_last_modelbool, default=False
save_init_modelbool, default=False
best_file_namestr, default=”best_model”
init_file_namestr, default=”init_model”
file_pathstr, default=”./”
stackslist of str or None, default=None

Stack types: “trend”, “seasonality”, or “generic”. Defaults to [“trend”, “seasonality”].

num_blocks_per_stackint, default=2
unitsint, default=30
num_trend_coefficientsint, default=3
num_seasonal_coefficientsint, default=5
num_generic_coefficientsint, default=7
share_weightsbool, default=True
share_coefficientsbool, default=True

Notes

Capabilities

Missing Values

No

Multithreading

No

Univariate

Yes

Multivariate

No

Horizon

Yes

Exogenous

No

References

[1]

Oreshkin et al. (2019). N-BEATS: Neural basis expansion analysis for interpretable time series forecasting. arXiv:1905.10437.

Examples

>>> from aeon.datasets import load_airline
>>> from aeon.forecasting.deep_learning import NBeatsForecaster
>>> y = load_airline()
>>> forecaster = NBeatsForecaster(window=24, horizon=12, n_epochs=10)
>>> forecaster.fit(y)
NBeatsForecaster(...)
>>> pred = forecaster.predict(y)

Methods

build_model(input_shape)

Build and compile the N-BEATS Keras model.

clone([random_state])

Obtain a clone of the object with the same hyperparameters.

fit(y[, exog, axis])

Fit forecaster to series y.

forecast(y[, exog, axis])

Forecast the next horizon steps ahead of y.

get_class_tag(tag_name[, raise_error, ...])

Get tag value from estimator class (only class tags).

get_class_tags()

Get class tags from estimator class and all its parent classes.

get_fitted_params([deep])

Get fitted parameters.

get_params([deep])

Get parameters for this estimator.

get_tag(tag_name[, raise_error, ...])

Get tag value from estimator class.

get_tags()

Get tags from estimator.

load_model(model_path)

Load a pre-trained keras model instead of fitting.

predict(y[, exog, axis])

Predict the next horizon steps ahead.

reset([keep])

Reset the object to a clean post-init state.

save_last_model_to_file([file_path])

Save the last epoch of the trained deep learning model.

series_to_series_forecast(y, prediction_horizon)

Forecast prediction_horizon using a series iterative approach.

set_fit_request(*[, axis, exog])

Configure whether metadata should be requested to be passed to the fit method.

set_params(**params)

Set the parameters of this estimator.

set_predict_request(*[, axis, exog])

Configure whether metadata should be requested to be passed to the predict method.

set_tags(**tag_dict)

Set dynamic tags to given values.

summary()

Summary function to return the losses/metrics for model fit.
build_model(input_shape)[source]

Build and compile the N-BEATS Keras model.

Declares both "forecast" (shape (batch, horizon)) and "backcast" (shape (batch, window)) as named outputs so the full graph is reachable and both heads receive direct gradients.

Parameters:
input_shapetuple

(window, 1) — channel dim satisfies NBeatsNetwork’s guard.

Returns:
modeltf.keras.Model
clone(random_state=None)[source]

Obtain a clone of the object with the same hyperparameters.

A clone is a different object without shared references, in post-init state. This function is equivalent to returning sklearn.clone of self. Equal in value to type(self)(**self.get_params(deep=False)).

Parameters:
random_stateint, RandomState instance, or None, default=None

Sets the random state of the clone. If None, the random state is not set. If int, random_state is the seed used by the random number generator. If RandomState instance, random_state is the random number generator.

Returns:
estimatorobject

Instance of type(self), clone of self (see above)

fit(y, exog=None, axis=1)[source]

Fit forecaster to series y.

Fit a forecaster to predict self.horizon steps ahead using y.

Parameters:
ynp.ndarray

A time series on which to learn a forecaster to predict horizon ahead.

exognp.ndarray, default =None

Optional target-time exogenous values aligned with y.

Returns:
self

Fitted BaseForecaster.

forecast(y, exog=None, axis=1) → float[source]

Forecast the next horizon steps ahead of y.

By default this is simply fit followed by predict.

Parameters:
ynp.ndarray

A time series to predict the next horizon value for. Must be of shape (n_channels, n_timepoints) if a multivariate time series.

exognp.ndarray, default =None

Optional target-time exogenous values aligned with y.

Returns:
float

single prediction self.horizon steps ahead of y.

classmethod get_class_tag(tag_name, raise_error=True, tag_value_default=None)[source]

Get tag value from estimator class (only class tags).

Parameters:
tag_namestr

Name of tag value.

raise_errorbool, default=True

Whether a ValueError is raised when the tag is not found.

tag_value_defaultany type, default=None

Default/fallback value if tag is not found and error is not raised.

Returns:
tag_value

Value of the tag_name tag in cls. If not found, returns an error if raise_error is True, otherwise it returns tag_value_default.

Raises:
ValueError

if raise_error is True and tag_name is not in self.get_tags().keys()

Examples

>>> from aeon.classification import DummyClassifier
>>> DummyClassifier.get_class_tag("capability:multivariate")
True
classmethod get_class_tags()[source]

Get class tags from estimator class and all its parent classes.

Returns:
collected_tagsdict

Dictionary of tag name and tag value pairs. Collected from _tags class attribute via nested inheritance. These are not overridden by dynamic tags set by set_tags or class __init__ calls.

get_fitted_params(deep=True)[source]

Get fitted parameters.

State required:

Requires state to be “fitted”.

Parameters:
deepbool, default=True

If True, will return the fitted parameters for this estimator and contained subobjects that are estimators.

Returns:
fitted_paramsdict

Fitted parameter names mapped to their values.

get_params(deep=True)

Get parameters for this estimator.

Parameters:
deepbool, default=True

If True, will return the parameters for this estimator and contained subobjects that are estimators.

Returns:
paramsdict

Parameter names mapped to their values.

get_tag(tag_name, raise_error=True, tag_value_default=None)[source]

Get tag value from estimator class.

Includes dynamic and overridden tags.

Parameters:
tag_namestr

Name of tag to be retrieved.

raise_errorbool, default=True

Whether a ValueError is raised when the tag is not found.

tag_value_defaultany type, default=None

Default/fallback value if tag is not found and error is not raised.

Returns:
tag_value

Value of the tag_name tag in self. If not found, returns an error if raise_error is True, otherwise it returns tag_value_default.

Raises:
ValueError

if raise_error is True and tag_name is not in self.get_tags().keys()

Examples

>>> from aeon.classification import DummyClassifier
>>> d = DummyClassifier()
>>> d.get_tag("capability:multivariate")
True
get_tags()[source]

Get tags from estimator.

Includes dynamic and overridden tags.

Returns:
collected_tagsdict

Dictionary of tag name and tag value pairs. Collected from _tags class attribute via nested inheritance and then any overridden and new tags from __init__ or set_tags.

load_model(model_path)[source]

Load a pre-trained keras model instead of fitting.

When calling this function, all functionalities can be used such as predict with the loaded model.

Parameters:
model_pathstr

Path to the saved model file including extension. Example: model_path=”path/to/file/best_model.keras”

Returns:
None
predict(y, exog=None, axis=1) → float[source]

Predict the next horizon steps ahead.

Parameters:
ynp.ndarray

A time series to predict the next horizon value for.

exognp.ndarray, default =None

Optional exogenous values for the prediction target. For models fitted with exogenous variables, this should contain the exogenous values needed to make the next prediction.

Returns:
float

single prediction self.horizon steps ahead of y.

reset(keep=None)[source]

Reset the object to a clean post-init state.

After a self.reset() call, self is equal or similar in value to type(self)(**self.get_params(deep=False)), assuming no other attributes were kept using keep.

Detailed behaviour:
removes any object attributes, except:

hyper-parameters (arguments of __init__) object attributes containing double-underscores, i.e., the string “__”

runs __init__ with current values of hyperparameters (result of get_params)

Not affected by the reset are:

object attributes containing double-underscores class and object methods, class attributes any attributes specified in the keep argument

Parameters:
keepNone, str, or list of str, default=None

If None, all attributes are removed except hyperparameters. If str, only the attribute with this name is kept. If list of str, only the attributes with these names are kept.

Returns:
selfobject

Reference to self.

Raises:
TypeError

If ‘keep’ is not a string or a list of strings.

save_last_model_to_file(file_path='./')[source]

Save the last epoch of the trained deep learning model.

Parameters:
file_pathstr, default=”./”

The directory where the model will be saved.

Returns:
None
series_to_series_forecast(y, prediction_horizon, exog=None) → ndarray[source]

Forecast prediction_horizon using a series iterative approach.

This function implements a series-to-series forecasting strategy. The forecaster is trained to predict multiple steps ahead in one go, returning a series of predictions. This is done by fitting the model once and then predicting a series of length prediction_horizon.

ynp.ndarray

The time series to make forecasts about. Must be of shape (n_channels, n_timepoints) if a multivariate time series.

prediction_horizonint

The number of future time steps to forecast.

exognp.ndarray, default =None

Optional exogenous time series data assumed to be aligned with y.

Returns:
np.ndarray

An array of shape (prediction_horizon,) containing the forecasts for each horizon.

Raises:
ValueError

if prediction_horizon less than 1.

set_fit_request(*, axis: bool | None | str = '$UNCHANGED$', exog: bool | None | str = '$UNCHANGED$') → NBeatsForecaster

Configure whether metadata should be requested to be passed to the fit method.

Note that this method is only relevant when this estimator is used as a sub-estimator within a meta-estimator and metadata routing is enabled with enable_metadata_routing=True (see sklearn.set_config). Please check the User Guide on how the routing mechanism works.

The options for each parameter are:

  • True: metadata is requested, and passed to fit if provided. The request is ignored if metadata is not provided.

  • False: metadata is not requested and the meta-estimator will not pass it to fit.

  • None: metadata is not requested, and the meta-estimator will raise an error if the user provides it.

  • str: metadata should be passed to the meta-estimator with this given alias instead of the original name.

The default (sklearn.utils.metadata_routing.UNCHANGED) retains the existing request. This allows you to change the request for some parameters and not others.

Added in version 1.3.

Parameters:
axisstr, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED

Metadata routing for axis parameter in fit.

exogstr, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED

Metadata routing for exog parameter in fit.

Returns:
selfobject

The updated object.

set_params(**params)

Set the parameters of this estimator.

The method works on simple estimators as well as on nested objects (such as Pipeline). The latter have parameters of the form <component>__<parameter> so that it’s possible to update each component of a nested object.

Parameters:
**paramsdict

Estimator parameters.

Returns:
selfestimator instance

Estimator instance.

set_predict_request(*, axis: bool | None | str = '$UNCHANGED$', exog: bool | None | str = '$UNCHANGED$') → NBeatsForecaster

Configure whether metadata should be requested to be passed to the predict method.

Note that this method is only relevant when this estimator is used as a sub-estimator within a meta-estimator and metadata routing is enabled with enable_metadata_routing=True (see sklearn.set_config). Please check the User Guide on how the routing mechanism works.

The options for each parameter are:

  • True: metadata is requested, and passed to predict if provided. The request is ignored if metadata is not provided.

  • False: metadata is not requested and the meta-estimator will not pass it to predict.

  • None: metadata is not requested, and the meta-estimator will raise an error if the user provides it.

  • str: metadata should be passed to the meta-estimator with this given alias instead of the original name.

The default (sklearn.utils.metadata_routing.UNCHANGED) retains the existing request. This allows you to change the request for some parameters and not others.

Added in version 1.3.

Parameters:
axisstr, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED

Metadata routing for axis parameter in predict.

exogstr, True, False, or None, default=sklearn.utils.metadata_routing.UNCHANGED

Metadata routing for exog parameter in predict.

Returns:
selfobject

The updated object.

set_tags(**tag_dict)[source]

Set dynamic tags to given values.

Parameters:
**tag_dictdict

Dictionary of tag name and tag value pairs.

Returns:
selfobject

Reference to self.

summary()[source]

Summary function to return the losses/metrics for model fit.

Returns:
historydict or None

Dictionary containing model’s train/validation losses and metrics.