finite_prediction_error#
- skfda.preprocessing.smoothing.validation.finite_prediction_error(hat_matrix)[source]#
Finite prediction error for cross validation.
\[\Xi(\nu,n) = \frac{1 + \frac{tr(\hat{H}^\nu)}{n}}{1 - \frac{tr(\hat{H}^\nu)}{n}}\]See [1] for additional details.
- Parameters:
hat_matrix (ndarray[Any, dtype[float64]]) – Smoothing matrix whose penalization score is desired.
- Returns:
Penalization given by the finite prediction error.
- Return type:
References