metrics

metrics

Functions:

Name	Description
validate_and_promote	Validate shape match and promote 1D to 2D.
rhos	Pearson correlation per dimension.
mean_rho	Mean Pearson correlation.
rmse	Root Mean Squared Error.
mae	Mean Absolute Error.

Attributes:

Name	Type	Description
MetricFunc	TypeAlias	MetricFunc is a function that takes (predictions, observations) and returns a metric value.

MetricFunc

MetricFunc: TypeAlias = Callable[[np.ndarray, np.ndarray], np.ndarray]

MetricFunc is a function that takes (predictions, observations) and returns a metric value.

validate_and_promote

validate_and_promote(predictions: np.ndarray, observations: np.ndarray) -> tuple[np.ndarray, np.ndarray]

Validate shape match and promote 1D to 2D.

rhos

rhos(predictions: np.ndarray, observations: np.ndarray) -> np.ndarray

Pearson correlation per dimension.

Parameters:

Name	Type	Description	Default
predictions	ndarray	(N,), (N, D), or (B, N, D).	required
observations	ndarray	Same shape as predictions.	required

Returns:

Type	Description
ndarray	(1,) for 1D input, (D,) for 2D, (B, D) for 3D.

mean_rho

mean_rho(predictions: np.ndarray, observations: np.ndarray) -> np.ndarray

Mean Pearson correlation.

Parameters:

Name	Type	Description	Default
predictions	ndarray	(N,), (N, D), or (B, N, D).	required
observations	ndarray	Same shape as predictions.	required

Returns:

Type	Description
ndarray	() for 1D/2D input, (B,) for 3D input.

rmse

rmse(predictions: np.ndarray, observations: np.ndarray) -> np.ndarray

Root Mean Squared Error.

Parameters:

Name	Type	Description	Default
predictions	ndarray	(N,), (N, D), or (B, N, D).	required
observations	ndarray	Same shape as predictions.	required

Returns:

Type	Description
ndarray	() for 1D/2D input, (B,) for 3D input.

mae

mae(predictions: np.ndarray, observations: np.ndarray) -> np.ndarray

Mean Absolute Error.

Parameters:

Name	Type	Description	Default
predictions	ndarray	(N,), (N, D), or (B, N, D).	required
observations	ndarray	Same shape as predictions.	required

Returns:

Type	Description
ndarray	() for 1D/2D input, (B,) for 3D input.