gluonts.model package#

class gluonts.model.Estimator(lead_time: int = 0, **kwargs)[source]#

Bases: object

An abstract class representing a trainable model.

The underlying model is trained by calling the train method with a training Dataset, producing a Predictor object.

classmethod derive_auto_fields(train_iter)[source]#
classmethod from_hyperparameters(**hyperparameters)[source]#
classmethod from_inputs(train_iter, **params)[source]#
lead_time: int#
prediction_length: int#
train(training_data: gluonts.dataset.Dataset, validation_data: Optional[gluonts.dataset.Dataset] = None) → gluonts.model.predictor.Predictor[source]#

Train the estimator on the given data.

Parameters

  • training_data – Dataset to train the model on.

  • validation_data – Dataset to validate the model on during training.

Returns

The predictor containing the trained model.

Return type

Predictor

class gluonts.model.Forecast[source]#

Bases: object

A abstract class representing predictions.

copy_aggregate(agg_fun: Callable)[source]#

Returns a new Forecast object with a time series aggregated over the dimension axis.

Parameters

agg_fun – Aggregation function that defines the aggregation operation (typically mean or sum).

copy_dim(dim: int)[source]#

Returns a new Forecast object with only the selected sub-dimension.

Parameters

dim – The returned forecast object will only represent this dimension.

dim() → int[source]#

Returns the dimensionality of the forecast object.

property freq#
property index: pandas.core.indexes.period.PeriodIndex#
info: Optional[Dict]#
item_id: Optional[str]#
mean: numpy.ndarray#
property median: numpy.ndarray#
plot(prediction_intervals=(50.0, 90.0), show_mean=False, color='b', label=None, output_file=None, *args, **kwargs)[source]#

Plots the median of the forecast as well as prediction interval bounds (requires matplotlib and pandas).

Parameters

  • prediction_intervals (float or list of floats in [0, 100]) – Prediction interval size(s). If a list, it will stack the error plots for each prediction interval. Only relevant for error styles with “ci” in the name.

  • show_mean (boolean) – Whether to also show the mean of the forecast.

  • color (matplotlib color name or dictionary) – The color used for plotting the forecast.

  • label (string) – A label (prefix) that is used for the forecast

  • output_file (str or None, default None) – Output path for the plot file. If None, plot is not saved to file.

  • args – Other arguments are passed to main plot() call

  • kwargs – Other keyword arguments are passed to main plot() call

prediction_length: int#
quantile(q: Union[float, str]) → numpy.ndarray[source]#

Computes a quantile from the predicted distribution.

Parameters

q – Quantile to compute.

Returns

Value of the quantile across the prediction range.

Return type

numpy.ndarray

quantile_ts(q: Union[float, str]) → pandas.core.series.Series[source]#
start_date: pandas._libs.tslibs.period.Period#
class gluonts.model.IncrementallyTrainable(*args, **kwds)[source]#

Bases: typing_extensions.Protocol

train_from(predictor: gluonts.model.predictor.Predictor, training_data: gluonts.dataset.Dataset, validation_data: Optional[gluonts.dataset.Dataset] = None) → gluonts.model.predictor.Predictor[source]#

Experimental: this feature may change in future versions. Train the estimator, starting from a previously trained predictor, on the given data.

Parameters

  • predictor – A previously trained model, from which to initialize the estimator training.

  • training_data – Dataset to train the model on.

  • validation_data – Dataset to validate the model on during training.

Returns

The predictor containing the trained model.

Return type

Predictor

class gluonts.model.Predictor(prediction_length: int, lead_time: int = 0)[source]#

Bases: object

Abstract class representing predictor objects. :param prediction_length: Prediction horizon.

classmethod derive_auto_fields(train_iter)[source]#
classmethod deserialize(path: pathlib.Path, **kwargs) → gluonts.model.predictor.Predictor[source]#

Load a serialized predictor from the given path.

Parameters

  • path – Path to the serialized files predictor.

  • **kwargs – Optional context/device parameter to be used with the predictor. If nothing is passed will use the GPU if available and CPU otherwise.

classmethod from_hyperparameters(**hyperparameters)[source]#
classmethod from_inputs(train_iter, **params)[source]#
predict(dataset: gluonts.dataset.Dataset, **kwargs) → Iterator[gluonts.model.forecast.Forecast][source]#

Compute forecasts for the time series in the provided dataset. This method is not implemented in this abstract class; please use one of the subclasses. :param dataset: The dataset containing the time series to predict.

Returns

Iterator over the forecasts, in the same order as the dataset iterable was provided.

Return type

Iterator[Forecast]

serialize(path: pathlib.Path) → None[source]#
class gluonts.model.QuantileForecast(forecast_arrays: numpy.ndarray, start_date: pandas._libs.tslibs.period.Period, forecast_keys: List[str], item_id: Optional[str] = None, info: Optional[Dict] = None)[source]#

Bases: gluonts.model.forecast.Forecast

A Forecast that contains arrays (i.e. time series) for quantiles and mean.

Parameters

  • forecast_arrays – An array of forecasts

  • start_date (pandas._libs.tslibs.period.Period) – start of the forecast

  • forecast_keys – A list of quantiles of the form ‘0.1’, ‘0.9’, etc., and potentially ‘mean’. Each entry corresponds to one array in forecast_arrays.

  • info (Optional[Dict]) – additional information that the forecaster may provide e.g. estimated parameters, number of iterations ran etc.

copy_dim(dim: int) → gluonts.model.forecast.QuantileForecast[source]#

Returns a new Forecast object with only the selected sub-dimension.

Parameters

dim – The returned forecast object will only represent this dimension.

dim() → int[source]#

Returns the dimensionality of the forecast object.

info: Optional[Dict]#
item_id: Optional[str]#
property mean: numpy.ndarray#

Forecast mean.

plot(label=None, output_file=None, keys=None, *args, **kwargs)[source]#

Plots the median of the forecast as well as prediction interval bounds (requires matplotlib and pandas).

Parameters

  • prediction_intervals (float or list of floats in [0, 100]) – Prediction interval size(s). If a list, it will stack the error plots for each prediction interval. Only relevant for error styles with “ci” in the name.

  • show_mean (boolean) – Whether to also show the mean of the forecast.

  • color (matplotlib color name or dictionary) – The color used for plotting the forecast.

  • label (string) – A label (prefix) that is used for the forecast

  • output_file (str or None, default None) – Output path for the plot file. If None, plot is not saved to file.

  • args – Other arguments are passed to main plot() call

  • kwargs – Other keyword arguments are passed to main plot() call

prediction_length: int#
quantile(inference_quantile: Union[float, str]) → numpy.ndarray[source]#

Computes a quantile from the predicted distribution.

Parameters

q – Quantile to compute.

Returns

Value of the quantile across the prediction range.

Return type

numpy.ndarray

start_date: pandas._libs.tslibs.period.Period#
class gluonts.model.SampleForecast(samples: numpy.ndarray, start_date: pandas._libs.tslibs.period.Period, item_id: Optional[str] = None, info: Optional[Dict] = None)[source]#

Bases: gluonts.model.forecast.Forecast

A Forecast object, where the predicted distribution is represented internally as samples.

Parameters

  • samples – Array of size (num_samples, prediction_length) (1D case) or (num_samples, prediction_length, target_dim) (multivariate case)

  • start_date (pandas._libs.tslibs.period.Period) – start of the forecast

  • info (Optional[Dict]) – additional information that the forecaster may provide e.g. estimated parameters, number of iterations ran etc.

copy_aggregate(agg_fun: Callable) → gluonts.model.forecast.SampleForecast[source]#

Returns a new Forecast object with a time series aggregated over the dimension axis.

Parameters

agg_fun – Aggregation function that defines the aggregation operation (typically mean or sum).

copy_dim(dim: int) → gluonts.model.forecast.SampleForecast[source]#

Returns a new Forecast object with only the selected sub-dimension.

Parameters

dim – The returned forecast object will only represent this dimension.

dim() → int[source]#

Returns the dimensionality of the forecast object.

info: Optional[Dict]#
item_id: Optional[str]#
property mean: numpy.ndarray#

Forecast mean.

property mean_ts: pandas.core.series.Series#

Forecast mean, as a pandas.Series object.

property num_samples#

The number of samples representing the forecast.

property prediction_length#

Time length of the forecast.

quantile(q: Union[float, str]) → numpy.ndarray[source]#

Computes a quantile from the predicted distribution.

Parameters

q – Quantile to compute.

Returns

Value of the quantile across the prediction range.

Return type

numpy.ndarray

start_date: pandas._libs.tslibs.period.Period#
to_quantile_forecast(quantiles: List[str]) → gluonts.model.forecast.QuantileForecast[source]#