kats.utils.parameter_tuning_utils module

Collection of methods that return default search spaces for their relevant models.

This module has a collection of functions. Each function is dedicated for a model. It returns default search space for hyperparameter tuning that pertains to the model. They are called by hyperparemeter tuning module, time_series_parameter_tuning.py.

Typical usage example:

SearchMethodFactory.create_search_method(get_default_prophet_parameter_search_space(), …) SearchMethodFactory.create_search_method(get_default_arnet_parameter_search_space(), …)

kats.utils.parameter_tuning_utils.get_default_arima_parameter_search_space() → List[Dict[str, Union[str, list, bool]]]

Generates default search space as a list of dictionaries and returns it for arima.

Each dictionary in the list corresponds to a hyperparameter, having properties defining that hyperparameter. Properties are name, type, value_type, values, is_ordered. Hyperparameters that are included: p, d, q.

Parameters

N/A –

Returns

As described above

Raises

N/A –

kats.utils.parameter_tuning_utils.get_default_arnet_parameter_search_space() → List[Dict[str, Union[str, list, bool]]]

Generates default search space as a list of dictionaries and returns it for arnet.

Each dictionary in the list corresponds to a hyperparameter, having properties defining that hyperparameter. Properties are name, type, value_type, values, is_ordered. Hyperparameters that are included: input_size, output_size, batch_size.

Parameters

N/A –

Returns

As described above

Raises

N/A –

kats.utils.parameter_tuning_utils.get_default_holtwinters_parameter_search_space() → List[Dict[str, Union[str, list, bool]]]

Generates default search space as a list of dictionaries and returns it for holtwinters.

Each dictionary in the list corresponds to a hyperparameter, having properties defining that hyperparameter. Properties are name, type, value_type, values, is_ordered. Hyperparameters that are included: trend, damped, seasonal, seasonal_periods.

Parameters

N/A –

Returns

As described above

Raises

N/A –

kats.utils.parameter_tuning_utils.get_default_prophet_parameter_search_space() → List[Dict[str, Union[str, list, bool]]]

Generates default search space as a list of dictionaries and returns it for prophet model.

Each dictionary in the list corresponds to a hyperparameter, having properties defining that hyperparameter. Properties are name, type, value_type, values, is_ordered. Hyperparameters that are included: seasonality_prior_scale, yearly_seasonality, weekly_seasonality, daily_seasonality, seasonality_mode, changepoint_prior_scale, changepoint_range.

Parameters

N/A –

Returns

As described above

Raises

N/A –

kats.utils.parameter_tuning_utils.get_default_sarima_parameter_search_space() → List[Dict[str, Union[str, list, bool]]]

Generates default search space as a list of dictionaries and returns it for sarima.

Each dictionary in the list corresponds to a hyperparameter, having properties defining that hyperparameter. Properties are name, type, value_type, values, is_ordered. Hyperparameters that are included: p, d, q, seasonal_order, trend.

Parameters

N/A –

Returns

As described above

Raises

N/A –

kats.utils.parameter_tuning_utils.get_default_stlf_parameter_search_space() → List[Dict[str, Union[str, list, bool]]]

Generates default search space as a list of dictionaries and returns it for stfl.

Each dictionary in the list corresponds to a hyperparameter, having properties defining that hyperparameter. Properties are name, type, value_type, values, is_ordered. Hyperparameters that are included: method, m.

Parameters

N/A –

Returns

As described above

Raises

N/A –

kats.utils.parameter_tuning_utils.get_default_theta_parameter_search_space() → List[Dict[str, Union[str, list, bool]]]

Generates default search space as a list of dictionaries and returns it for theta.

Each dictionary in the list corresponds to a hyperparameter, having properties defining that hyperparameter. Properties are name, type, value_type, values, is_ordered. Hyperparameters that are included: m.

Parameters

N/A –

Returns

As described above

Raises

N/A –

kats.utils.parameter_tuning_utils.get_default_var_parameter_search_space() → List[Dict[str, Union[str, list, bool]]]

Generates default search space as a list of dictionaries and returns it for var.

Each dictionary in the list corresponds to a hyperparameter, having properties defining that hyperparameter. Properties are name, type, value_type, values, is_ordered. Hyperparameters that are included: To be filled.

Parameters

N/A –

Returns

As described above

Raises

• NotImplementedError – The method is to be implemented in the future. This

• error will then be removed. –