Skip to content

Statsmodels Adapters API

This section documents adapters for integrating classic statsmodels time-series models into the eb-adapters package.

All content below is generated automatically from NumPy-style docstrings in the source code.

Statsmodels Adapter Module

eb_adapters.statsmodels

SarimaxAdapter

Bases: BaseAdapter

Adapter for statsmodels SARIMAX.

This wrapper fits a univariate SARIMAX model on y and produces forecasts for len(X) steps ahead when predict(X) is called.

Parameters:

Name Type Description Default
order tuple[int, int, int]

ARIMA (p, d, q) order.

(1, 0, 0)
seasonal_order tuple[int, int, int, int]

Seasonal (P, D, Q, s) order.

(0, 0, 0, 0)
trend str | None

Trend specification forwarded to SARIMAX.

None
enforce_stationarity bool

Whether to enforce stationarity in the SARIMAX model.

True
enforce_invertibility bool

Whether to enforce invertibility in the SARIMAX model.

True
Notes
  • X is ignored during fitting. It is only used at prediction time to determine the forecast horizon (n_steps = len(X)).
  • This adapter stores initialization parameters via get_params() so cloning utilities can reconstruct the adapter.

fit(X, y, sample_weight=None)

Fit a univariate SARIMAX model on y.

Parameters:

Name Type Description Default
X ndarray

Ignored. Present for API compatibility.

required
y ndarray

Target series of shape (n_samples,).

required
sample_weight ndarray | None

Accepted for API compatibility but ignored by this adapter.

None

Returns:

Type Description
SarimaxAdapter

The fitted adapter (self), allowing method chaining.

Raises:

Type Description
ImportError

If statsmodels is not installed.

predict(X)

Forecast len(X) steps ahead from the end of the training sample.

Parameters:

Name Type Description Default
X ndarray

Array-like placeholder used to determine the forecast horizon.

required

Returns:

Type Description
ndarray

Forecast values of shape (len(X),).

Raises:

Type Description
RuntimeError

If the adapter has not been fit yet.

get_params(deep=True)

Return initialization parameters for cloning utilities.

Parameters:

Name Type Description Default
deep bool

Included for scikit-learn compatibility.

True

Returns:

Type Description
dict[str, Any]

Initialization parameters that can be passed back to __init__.

set_params(**params)

Update adapter parameters.

Parameters:

Name Type Description Default
**params Any

Parameters to set as attributes on the adapter instance.

{}

Returns:

Type Description
SarimaxAdapter

The updated adapter instance (self).

ArimaAdapter

Bases: BaseAdapter

Adapter for statsmodels ARIMA.

This wrapper fits a univariate ARIMA model on y and produces forecasts for len(X) steps ahead when predict(X) is called.

Parameters:

Name Type Description Default
order tuple[int, int, int]

ARIMA (p, d, q) order.

(1, 0, 0)
trend str | None

Trend specification forwarded to statsmodels.tsa.ARIMA.

None
Notes
  • X is ignored during fitting. It is only used at prediction time to determine the forecast horizon (n_steps = len(X)).
  • This adapter stores initialization parameters via get_params() so cloning utilities can reconstruct the adapter.

fit(X, y, sample_weight=None)

Fit a univariate ARIMA model on y.

Parameters:

Name Type Description Default
X ndarray

Ignored. Present for API compatibility.

required
y ndarray

Target series of shape (n_samples,).

required
sample_weight ndarray | None

Accepted for API compatibility but ignored by this adapter.

None

Returns:

Type Description
ArimaAdapter

The fitted adapter (self), allowing method chaining.

Raises:

Type Description
ImportError

If statsmodels is not installed.

predict(X)

Forecast len(X) steps ahead from the end of the training sample.

Parameters:

Name Type Description Default
X ndarray

Array-like placeholder used to determine the forecast horizon.

required

Returns:

Type Description
ndarray

Forecast values of shape (len(X),).

Raises:

Type Description
RuntimeError

If the adapter has not been fit yet.

get_params(deep=True)

Return initialization parameters for cloning utilities.

Parameters:

Name Type Description Default
deep bool

Included for scikit-learn compatibility.

True

Returns:

Type Description
dict[str, Any]

Initialization parameters that can be passed back to __init__.

set_params(**params)

Update adapter parameters.

Parameters:

Name Type Description Default
**params Any

Parameters to set as attributes on the adapter instance.

{}

Returns:

Type Description
ArimaAdapter

The updated adapter instance (self).