Autoformer

chronax.models.autoformer.forecaster · inherits BaseForecaster

Univariate Autoformer forecaster (JAX/Flax Linen port of neuralforecast.Autoformer).

Maintenance status: Active univariate forecaster. Integrates with the BaseForecaster interface, including conformal prediction intervals via predict(level=...). Defaults match neuralforecast.Autoformer.

Attributes

Attribute	Type	Description
uses_exog	bool	False
alias	str	Alias used for logging and identification.
conformal_params	ConformalIntervals | None	Parameters used for conformal prediction intervals.
model_	TrainState | None	The fitted Flax training state.

__init__(self, h: int, input_size: int = -1, hidden_size: int = 128, n_heads: int = 4, factor: int = 3, moving_avg_window: int = 25, encoder_layers: int = 2, decoder_layers: int = 1, conv_hidden_size: int = 32, decoder_input_size_multiplier: float = 0.5, dropout: float = 0.05, activation: str = 'gelu', max_steps: int = 1000, learning_rate: float = 0.0001, batch_size: int = 32, num_lr_decays: int = 3, val_fraction: float = 0.1, val_check_steps: int = 100, early_stop_patience_steps: int = -1, grad_clip: float = 1.0, weight_decay: float = 0.0, random_seed: int = 1, alias: str = 'Autoformer', loss: Union[str, LossFn] = 'mae')

(undocumented)

Parameter	Type	Default	Description
h	int	-	(undocumented)
input_size	int	-1	(undocumented)
hidden_size	int	128	(undocumented)
n_heads	int	4	(undocumented)
factor	int	3	(undocumented)
moving_avg_window	int	25	(undocumented)
encoder_layers	int	2	(undocumented)
decoder_layers	int	1	(undocumented)
conv_hidden_size	int	32	(undocumented)
decoder_input_size_multiplier	float	0.5	(undocumented)
dropout	float	0.05	(undocumented)
activation	str	'gelu'	(undocumented)
max_steps	int	1000	(undocumented)
learning_rate	float	0.0001	(undocumented)
batch_size	int	32	(undocumented)
num_lr_decays	int	3	(undocumented)
val_fraction	float	0.1	(undocumented)
val_check_steps	int	100	(undocumented)
early_stop_patience_steps	int	-1	(undocumented)
grad_clip	float	1.0	(undocumented)
weight_decay	float	0.0	(undocumented)
random_seed	int	1	(undocumented)
alias	str	'Autoformer'	(undocumented)
loss	Union[str, LossFn]	'mae'	(undocumented)

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

Fit on a univariate 1-D series.

Parameters:

Parameter	Type	Default	Description
y	jnp.ndarray	-	(undocumented)
X	jnp.ndarray | None	None	(undocumented)

Returns: Self (the fitted forecaster; sets self.model_). Raises: NotImplementedError (if X is not None), ValueError (if y is not 1-D or too short).

predict(self, h: int, X: jnp.ndarray | None = None, level: list[int | float] | None = None) -> dict

Forecast h steps from the fitted context.

Parameters:

Parameter	Type	Default	Description
h	int	-	(undocumented)
X	jnp.ndarray | None	None	(undocumented)
level	list[int | float] | None	None	(undocumented)

Returns: dict containing the forecast.

Key	Type	Description
"mean"	jnp.ndarray	The point forecast of shape (h,).
lower_<level>	jnp.ndarray	Lower bound of the prediction interval (if level is provided).
upper_<level>	jnp.ndarray	Upper bound of the prediction interval (if level is provided).

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

Stateless fit-then-predict on y.

Parameters:

Parameter	Type	Default	Description
y	jnp.ndarray	-	(undocumented)
h	int	-	(undocumented)
X	jnp.ndarray | None	None	(undocumented)
X_future	jnp.ndarray | None	None	(undocumented)
level	list[int | float] | None	None	(undocumented)
fitted	bool	False	(undocumented)

Returns: dict containing the forecast.

Key	Type	Description
"mean"	jnp.ndarray	The point forecast.
"fitted"	jnp.ndarray	One-step-ahead fitted values (if fitted=True).
lower_<level>	jnp.ndarray	Lower bound of the prediction interval (if level is provided).
upper_<level>	jnp.ndarray	Upper bound of the prediction interval (if level is provided).
Raises: NotImplementedError (if X or X_future is not None).

AutoformerForecaster

chronax.models.autoformer.forecaster · inherits Autoformer

Deprecated config-based constructor; prefer :class:Autoformer.

__init__(self, config: AutoformerConfig, *, max_steps: int = 1000, learning_rate: float = 0.0001, batch_size: int = 32, num_lr_decays: int = 3, val_fraction: float = 0.1, val_check_steps: int = 100, early_stop_patience_steps: int = -1, grad_clip: float = 1.0, weight_decay: float = 0.0, loss: Union[str, LossFn, Callable] = 'mae', seed: int = 1)

Deprecated constructor. Use Autoformer(h=..., input_size=..., random_seed=...) instead.

Parameter	Type	Default	Description
config	AutoformerConfig	-	(undocumented)
max_steps	int	1000	(undocumented)
learning_rate	float	0.0001	(undocumented)
batch_size	int	32	(undocumented)
num_lr_decays	int	3	(undocumented)
val_fraction	float	0.1	(undocumented)
val_check_steps	int	100	(undocumented)
early_stop_patience_steps	int	-1	(undocumented)
grad_clip	float	1.0	(undocumented)
weight_decay	float	0.0	(undocumented)
loss	Union[str, LossFn, Callable]	'mae'	(undocumented)
seed	int	1	(undocumented)