RandomWalkWithDrift

chronax.models.RandomWalkWithDrift · inherits BaseForecaster

The RandomWalkWithDrift class implements statsforecast's random walk with drift forecasting model. A variation of the naive method that allows forecasts to change over time by extrapolating a linear trend between the first and last observations. This JAX implementation provides complete compatibility with statsforecast's RandomWalkWithDrift class.

__init__(self, alias: str = 'RWD', conformal_params: ConformalIntervals | None = None)

Initializes the RandomWalkWithDrift model.

Parameter	Type	Default	Description
alias	str	"RWD"	(undocumented)
conformal_params	ConformalIntervals | None	None	(undocumented)

fit(self, y: jnp.ndarray, X: jnp.ndarray | None = None) -> Self

Fit the RandomWalkWithDrift model.

Parameters:

Parameter	Type	Default	Description
y	jnp.ndarray	-	Clean time series of shape (t,)
X	jnp.ndarray | None	None	Optional exogenous variables (not used, for API compatibility)

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

predict(self, h: int, level: list[int] | None = None) -> dict

Predict with fitted RandomWalkWithDrift.

Parameters:

Parameter	Type	Default	Description
h	int	-	Forecast horizon
level	list[int] | None	None	Confidence levels (0-100) for prediction intervals

Returns: dict (Dictionary with entries mean for point predictions and level_* for probabilistic predictions). Keys include {"mean": jnp.ndarray}. If level is provided, keys also include lo-L and hi-L for each level L.

predict_in_sample(self, level: list[int] | None = None) -> dict

Access fitted RandomWalkWithDrift insample predictions.

Parameters:

Parameter	Type	Default	Description
level	list[int] | None	None	Confidence levels (0-100) for prediction intervals

Returns: dict (Dictionary with entries fitted for point predictions). Keys include {"fitted": jnp.ndarray}. If level is provided, keys also include fitted-lo-L and fitted-hi-L for each level L.

forecast(self, h: int, y: jnp.ndarray, X: jnp.ndarray | None = None, X_future: jnp.ndarray | None = None, level: list[int] | None = None, fitted: bool = False) -> dict

Memory Efficient RandomWalkWithDrift predictions.

This method avoids memory burden due from object storage. It is analogous to fit_predict without storing information. It assumes you know the forecast horizon in advance.

Parameters:

Parameter	Type	Default	Description
h	int	-	Forecast horizon
y	jnp.ndarray	-	Clean time series of shape (n,)
X	jnp.ndarray | None	None	Optional insample exogenous of shape (t, n_x) (not used, for API compatibility)
X_future	jnp.ndarray | None	None	Optional exogenous of shape (h, n_x) (not used, for API compatibility)
level	list[int] | None	None	Confidence levels (0-100) for prediction intervals
fitted	bool	False	Whether or not to return insample predictions

Returns: dict (Dictionary with entries mean for point predictions and level_* for probabilistic predictions). Keys include {"mean": jnp.ndarray}. If fitted=True, includes "fitted". If level is provided, includes interval keys (lo-L, hi-L, and optionally fitted-lo-L, fitted-hi-L).