actableai.timeseries.models.params package

Submodules

actableai.timeseries.models.params.base module

class actableai.timeseries.models.params.base.BaseParams(model_name: str, is_multivariate_model: bool, has_estimator: bool = True, handle_feat_static_real: bool = True, handle_feat_static_cat: bool = True, handle_feat_dynamic_real: bool = False, handle_feat_dynamic_cat: bool = False, hyperparameters: Optional[Dict] = None, process_hyperparameters: bool = True)

Bases: object

Base class for Time Series Model Parameters.

build_estimator(*, ctx: Context, freq: str, prediction_length: int, target_dim: int, distr_output: DistributionOutput, params: Dict[str, Any]) → Optional[AAITimeSeriesEstimator]

Build an estimator from the underlying model using selected parameters.

Parameters

• ctx – mxnet context.
• freq – Frequency of the time series used.
• prediction_length – Length of the prediction that will be forecasted.
• target_dim – Target dimension (number of columns to predict).
• distr_output – Distribution output to use.
• params – Selected parameters from the hyperparameter space.

Returns

Built estimator.

build_predictor(*, freq: str, prediction_length: int, target_dim: int, params: Dict[str, Any]) → Optional[AAITimeSeriesPredictor]

Build a predictor from the underlying model using selected parameters.

Parameters

• freq – Frequency of the time series used.
• prediction_length – Length of the prediction that will be forecasted.
• target_dim – Target dimension (number of columns to predict).
• params – Selected parameters from the hyperparameter space.

Returns

Built predictor.

static get_hyperparameters() → Parameters

Returns the hyperparameters space of the model.

Returns

The hyperparameters space.

setup(use_feat_static_real: bool, use_feat_static_cat: bool, use_feat_dynamic_real: bool, use_feat_dynamic_cat: bool)

Set up the parameters.

Parameters

• use_feat_static_real – True if the data contains real static features.
• use_feat_static_cat – True if the data contains categorical static features.
• use_feat_dynamic_real – True if the data contains real dynamic features.
• use_feat_dynamic_cat – True if the data contains categorical dynamic features.

tune_config(prediction_length: int) → Dict[str, Any]

Select parameters in the pre-defined hyperparameter space.

Parameters

prediction_length – Length of the prediction.

Returns

Selected parameters.

class actableai.timeseries.models.params.base.Model(value)

Bases: str, enum.Enum

Enum representing the different models available.

constant_value = 'constant_value'

deep_ar = 'deep_ar'

deep_var = 'deep_var'

feed_forward = 'feed_forward'

gp_var = 'gp_var'

multivariate_constant_value = 'multivariate_constant_value'

n_beats = 'n_beats'

prophet = 'prophet'

r_forecast = 'r_forecast'

tree_predictor = 'tree_predictor'

actableai.timeseries.models.params.constant_value module

class actableai.timeseries.models.params.constant_value.ConstantValueParams(hyperparameters: Optional[Dict] = None, process_hyperparameters: bool = True, multivariate: bool = False)

Bases: actableai.timeseries.models.params.base.BaseParams

Parameters class for the Constant Value Model.

build_predictor(*, prediction_length: int, target_dim: int, params: Dict[str, Any], **kwargs) → AAITimeSeriesPredictor

Build a predictor from the underlying model using selected parameters.

Parameters

• prediction_length – Length of the prediction that will be forecasted.
• target_dim – Target dimension (number of columns to predict).
• params – Selected parameters from the hyperparameter space.
• kwargs – Ignored arguments.

Returns

Built predictor.

static get_hyperparameters() → Parameters

Returns the hyperparameters space of the model.

Returns

The hyperparameters space.

tune_config(prediction_length) → Dict[str, Any]

Select parameters in the pre-defined hyperparameter space.

Returns

Selected parameters.

class actableai.timeseries.models.params.constant_value.MultivariateConstantValueParams(hyperparameters: Optional[Dict] = None, process_hyperparameters: bool = True)

Bases: actableai.timeseries.models.params.constant_value.ConstantValueParams

Parameters class for the Multivariate Constant Value Model.

static get_hyperparameters() → Parameters

Returns the hyperparameters space of the model.

Returns

The hyperparameters space.

actableai.timeseries.models.params.deep_ar module

class actableai.timeseries.models.params.deep_ar.DeepARParams(hyperparameters: Optional[Dict] = None, process_hyperparameters: bool = True)

Bases: actableai.timeseries.models.params.base.BaseParams

Parameter class for Deep AR Model.

build_estimator(*, ctx: Context, freq: str, prediction_length: int, distr_output: DistributionOutput, params: Dict[str, Any], **kwargs) → AAITimeSeriesEstimator

Build an estimator from the underlying model using selected parameters.

Parameters

• ctx – mxnet context.
• freq – Frequency of the time series used.
• prediction_length – Length of the prediction that will be forecasted
• distr_output – Distribution output to use.
• params – Selected parameters from the hyperparameter space.
• kwargs – Ignored arguments.

Returns

Built estimator.

static get_hyperparameters() → Parameters

Returns the hyperparameters space of the model.

Returns

The hyperparameters space.

tune_config(prediction_length) → Dict[str, Any]

Select parameters in the pre-defined hyperparameter space.

Returns

Selected parameters.

actableai.timeseries.models.params.deep_var module

class actableai.timeseries.models.params.deep_var.DeepVARParams(hyperparameters: Optional[Dict] = None, process_hyperparameters: bool = True)

Bases: actableai.timeseries.models.params.base.BaseParams

Parameter class for Deep VAR Model.

build_estimator(*, ctx: Context, freq: str, prediction_length: int, target_dim: int, params: Dict[str, Any], **kwargs) → AAITimeSeriesEstimator

Build an estimator from the underlying model using selected parameters.

Parameters

• ctx – mxnet context.
• freq – Frequency of the time series used.
• prediction_length – Length of the prediction that will be forecasted.
• target_dim – Target dimension (number of columns to predict).
• params – Selected parameters from the hyperparameter space.
• kwargs – Ignored arguments.

Returns

Built estimator.

static get_hyperparameters() → Parameters

Returns the hyperparameters space of the model.

Returns

The hyperparameters space.

tune_config(prediction_length) → Dict[str, Any]

Select parameters in the pre-defined hyperparameter space.

Returns

Selected parameters.

actableai.timeseries.models.params.feed_forward module

class actableai.timeseries.models.params.feed_forward.FeedForwardParams(hyperparameters: Optional[Dict] = None, process_hyperparameters: bool = True)

Bases: actableai.timeseries.models.params.base.BaseParams

Parameter class for Feed Forward Model.

build_estimator(*, ctx: Context, prediction_length: int, distr_output: DistributionOutput, params: Dict[str, Any], **kwargs) → AAITimeSeriesEstimator

Build an estimator from the underlying model using selected parameters.

Parameters

• ctx – mxnet context.
• prediction_length – Length of the prediction that will be forecasted.
• distr_output – Distribution output to use.
• params – Selected parameters from the hyperparameter space.
• kwargs – Ignored arguments.

Returns

Built estimator.

static get_hyperparameters() → Parameters

Returns the hyperparameters space of the model.

Returns

The hyperparameters space.

tune_config(prediction_length) → Dict[str, Any]

Select parameters in the pre-defined hyperparameter space.

Returns

Selected parameters.

actableai.timeseries.models.params.gp_var module

class actableai.timeseries.models.params.gp_var.GPVarParams(hyperparameters: Optional[Dict] = None, process_hyperparameters: bool = True)

Bases: actableai.timeseries.models.params.base.BaseParams

Parameter class for GP Var Model.

build_estimator(*, ctx: Context, freq: str, prediction_length: int, target_dim: int, params: Dict[str, Any], **kwargs) → AAITimeSeriesEstimator

Build an estimator from the underlying model using selected parameters.

Parameters

• ctx – mxnet context.
• freq – Frequency of the time series used.
• prediction_length – Length of the prediction that will be forecasted.
• target_dim – Target dimension (number of columns to predict).
• params – Selected parameters from the hyperparameter space.
• kwargs – Ignored arguments.

Returns

Built estimator.

static get_hyperparameters() → Parameters

Returns the hyperparameters space of the model.

Returns

The hyperparameters space.

tune_config(prediction_length) → Dict[str, Any]

Select parameters in the pre-defined hyperparameter space.

Returns

Selected parameters.

actableai.timeseries.models.params.n_beats module

class actableai.timeseries.models.params.n_beats.NBEATSParams(hyperparameters: Optional[Dict] = None, process_hyperparameters: bool = True)

Bases: actableai.timeseries.models.params.base.BaseParams

Parameter class for NBEATS Model.

build_estimator(*, ctx: Context, freq: str, prediction_length: int, params: Dict[str, Any], **kwargs) → AAITimeSeriesEstimator

Build an estimator from the underlying model using selected parameters.

Parameters

• ctx – mxnet context.
• freq – Frequency of the time series used.
• prediction_length – Length of the prediction that will be forecasted.
• params – Selected parameters from the hyperparameter space.
• kwargs – Ignored arguments.

Returns

Built estimator.

static get_hyperparameters() → Parameters

Returns the hyperparameters space of the model.

Returns

The hyperparameters space.

tune_config(prediction_length) → Dict[str, Any]

Select parameters in the pre-defined hyperparameter space.

Returns

Selected parameters.

actableai.timeseries.models.params.prophet module

class actableai.timeseries.models.params.prophet.ProphetParams(hyperparameters: Optional[Dict] = None, process_hyperparameters: bool = True)

Bases: actableai.timeseries.models.params.base.BaseParams

Parameter class for Prophet Model.

build_predictor(*, prediction_length: int, params: Dict[str, Any], **kwargs) → AAITimeSeriesPredictor

Build a predictor from the underlying model using selected parameters.

Parameters

• prediction_length – Length of the prediction that will be forecasted.
• params – Selected parameters from the hyperparameter space.
• kwargs – Ignored arguments.

Returns

Built predictor.

static get_hyperparameters() → Parameters

Returns the hyperparameters space of the model.

Returns

The hyperparameters space.

tune_config(prediction_length) → Dict[str, Any]

Select parameters in the pre-defined hyperparameter space.

Returns

Selected parameters.

actableai.timeseries.models.params.r_forecast module

class actableai.timeseries.models.params.r_forecast.RForecastParams(hyperparameters: Optional[Dict] = None, process_hyperparameters: bool = True)

Bases: actableai.timeseries.models.params.base.BaseParams

Parameters class for RForecast Model.

build_predictor(*, freq: str, prediction_length: int, params: Dict[str, Any], **kwargs) → AAITimeSeriesPredictor

Build a predictor from the underlying model using selected parameters.

Parameters

• freq – Frequency of the time series used.
• prediction_length – Length of the prediction that will be forecasted.
• params – Selected parameters from the hyperparameter space.
• kwargs – Ignored arguments.

Returns

Built predictor.

static get_hyperparameters() → Parameters

Returns the hyperparameters space of the model.

Returns

The hyperparameters space.

tune_config(prediction_length) → Dict[str, Any]

Select parameters in the pre-defined hyperparameter space.

Returns

Selected parameters.

actableai.timeseries.models.params.tree_predictor module

class actableai.timeseries.models.params.tree_predictor.TreePredictorParams(hyperparameters: Optional[Dict] = None, process_hyperparameters: bool = True)

Bases: actableai.timeseries.models.params.base.BaseParams

Parameters class for Tree Predictor Model.

build_estimator(*, freq: str, prediction_length: int, params: Dict[str, Any], **kwargs) → AAITimeSeriesEstimator

Build an estimator from the underlying model using selected parameters.

Parameters

• freq – Frequency of the time series used.
• prediction_length – Length of the prediction that will be forecasted.
• params – Selected parameters from the hyperparameter space.
• kwargs – Ignored arguments.

Returns

Built estimator.

static get_hyperparameters() → Parameters

Returns the hyperparameters space of the model.

Returns

The hyperparameters space.

tune_config(prediction_length) → Dict[str, Any]

Select parameters in the pre-defined hyperparameter space.

Returns

Selected parameters.

Module contents