reid#

hidimstat.desparsified_lasso.reid(beta_hat, residual, tolerance=0.0001, multioutput=False, stationary=True, method='median', order=1)[source]#

Residual sum of squares based estimators for noise standard deviation estimation.

This implementation follows the procedure described in Fan et al.[1] and Reid et al.[2]. The beta_hat should correspond to the coefficient of Lasso with cross-validation, and the residual is based on this model.

For group, the implementation is based on the procedure from Chevalier et al.[3].

Parameters:
beta_hatndarray, shape (n_features,) or (n_task, n_features)

Estimated sparse coefficient vector from regression.

residualndarray, shape (n_samples,) or (n_samples, n_task)

Residuals from the regression model.

tolerancefloat, default=1e-4

Threshold for considering coefficients as non-zero.

multioutputbool, default=False

If True, handles multiple outputs (group case).

stationarybool, default=True

Whether noise has constant magnitude across time steps.

method{‘median’, ‘AR’}, (default=’simple’)

Covariance estimation method: - ‘median’: Uses median correlation between consecutive time steps - ‘AR’: Uses Yule-Walker method with specified order

orderint, default=1

Order of AR model when method=’AR’. Must be < n_task.

Returns:
sigma_hat_raw or covariance_hatfloat or ndarray

For single output: estimated noise standard deviation For multiple outputs: estimated (n_task, n_task) covariance matrix

Notes

Implementation based on Reid et al.[2] for single output and Chevalier et al.[3] for multiple outputs.

References