ahat.compute_linear_pod_curve

compute_linear_pod_curve(X_eval, model, tau, threshold, xlog=False, ylog=False)

Calculates the PoD curve using the standard a-hat vs a analytical assumptions.

Assumes that the residual errors are perfectly normally distributed with a constant standard deviation (tau).

Parameters

Name	Type	Description	Default
X_eval	np.ndarray	The 1D grid of points to evaluate the PoD curve.	required
model	LinearRegression	The fitted linear expectation model.	required
tau	float	The constant standard deviation of the residuals.	required
threshold	float	The detection threshold in original (untransformed) units.	required
xlog	bool	Indicates if the model was trained with log-transformed X.	False
ylog	bool	Indicates if the model was trained with log-transformed y.	False

Returns

Name	Type	Description
	Tuple[np.ndarray, np.ndarray]	Tuple[np.ndarray, np.ndarray]: - pod_curve: Array of probabilities [0, 1] for each point in X_eval. - mean_curve: Array of expected mean signal responses in original units.

Examples

# Assuming we have the model and tau from the fitting step
X_eval = np.linspace(1, 10, 100)
pod, mean_response = compute_linear_pod_curve(
    X_eval, model, tau, threshold=3.5, xlog=True, ylog=False
)