Cp = (RSS + 2 d sigma^2) / n (ESL eq 7.26; for a linear smoother d =
trace(S), eq 6.35), with the classical RSS / sigma^2 - n + 2 d alongside.
Usage
morie_esl_mallows_cp(RSS, d, n, sigma2)
Arguments
Residual sum of squares.
- d
Number of parameters or effective degrees of freedom.
- n
Sample size.
- sigma2
Noise variance estimate.
Value
Named list: estimate, cp_classical, RSS, d, n, sigma2.
References
Hastie, Tibshirani & Friedman (2009), sec. 7.5.
Examples
morie_esl_mallows_cp(10, 3, 50, 0.5)$estimate
#> [1] 0.26
