SimpleExponentialSmoothing

chronax.models.SimpleExponentialSmoothing · inherits BaseForecaster

JAX-optimized Simple Exponential Smoothing.

Weighted average of past observations with exponentially decreasing weights. Formula: $\hat{y}_{t+1} = \alpha \cdot y_t + (1-\alpha) \cdot \hat{y}_t$

__init__(self, alpha, alias='SES', conformal_params=None)

Initializes the Simple Exponential Smoothing model parameters.

Parameter	Type	Default	Description
alpha	float	-	Smoothing parameter in [0,1]
alias	str	"SES"	Model name
conformal_params	ConformalIntervals \| None	None	ConformalIntervals for prediction intervals

fit(self, y, X=None) -> Self

Fit the SimpleExponentialSmoothing model.

Runs SES on the full series to produce fitted values and the final smoothed level (used as the flat forecast). If conformal_params is configured, conformity scores are computed and cached for predict().

Parameter	Type	Default	Description
y	jnp.ndarray	-	Clean time series of shape (t,).
X	jnp.ndarray \| None	None	Exogenous variables (unused; included for API compatibility).

Returns: Self (the fitted forecaster; sets self.model_).

forecast(self, y, h, X=None, X_future=None, level=None, fitted=False) -> dict

Memory-efficient stateless fit+predict in one call.

Fits SES on y and immediately generates h-step ahead point forecasts without storing any model state. Used internally by BaseForecaster for conformity score computation in cross-validation windows.

Parameter	Type	Default	Description
y	jnp.ndarray	-	Clean time series of shape (t,).
h	int	-	Forecast horizon (number of steps ahead).
X	jnp.ndarray \| None	None	In-sample exogenous variables (unused; included for API compatibility).
X_future	jnp.ndarray \| None	None	Future exogenous variables (unused; included for API compatibility).
level	list[int \| float] \| None	None	Confidence levels (unused; included for BaseForecaster compliance).
fitted	bool	False	Whether to return fitted values (unused; included for BaseForecaster compliance).

Returns: dict: Dictionary containing "mean", point forecasts of shape (h,), all equal to the final smoothed level. * "mean": jnp.ndarray (shape (h,))

predict(self, h, X=None, level=None) -> dict

Generate h-step ahead forecasts using the fitted model.

All h forecasts equal the final smoothed level $\ell[T]$. Optionally adds conformal prediction intervals using cached conformity scores from fit().

Parameter	Type	Default	Description
h	int	-	Forecast horizon (number of steps ahead).
X	jnp.ndarray \| None	None	Exogenous variables (unused; included for API compatibility).
level	list[int] \| None	None	Confidence levels (0--100) for prediction intervals, e.g. [80, 95]. Requires conformal_params to be set.

Returns: dict: Dictionary containing "mean" (point forecasts of shape (h,)) and optionally "lo-{l}" / "hi-{l}" (conformal interval bounds for each level l, only present when level is not None). * "mean": jnp.ndarray (shape (h,)) * "lo-{l}": jnp.ndarray (shape (h,), if level is provided) * "hi-{l}": jnp.ndarray (shape (h,), if level is provided)

Raises: * ValueError: If level is requested but conformal_params is None. * ValueError: If level is requested but the model has not been fitted yet.

predict_in_sample(self) -> dict

Return in-sample fitted values from the last fit() call.

Parameter	Type	Default	Description
self	-	-	(undocumented)

Returns: dict: Dictionary containing: * "fitted": In-sample smoothed predictions of shape (t,). The first value is NaN (no prior level available at t=0).

Raises: * ValueError: If the model has not been fitted yet.