CrossValidatedSmoothing Class Reference

Determines the spline smoothing factor as in Clark (1977) More...

#include <SmoothingSpline.h>

Public Member Functions
void	set_spline (SmoothingSpline *_spline)
void	fit (std::vector< double > &data_x, std::vector< Estimate< double > > &data_y)
	Fit spline to data using current configuration.
double	get_mean_gof (const std::vector< double > &data_x, const std::vector< Estimate< double > > &data_y)
	Return the mean goodness-of-fit for the current smoothing.
void	get_nfree_trials (std::vector< double > &nfree, unsigned ndat)
	Get the trial smoothing factors.

Detailed Description

Determines the spline smoothing factor as in Clark (1977)

For small numbers of data points to be fit and at low signal-to-noise ratios, the GCV function exhibits multiple local minima and in practice, for around 10% to 20% of trials, the GCVSPL sub-routine yields smoothing splines that overfit the data.

The m-fold cross-validation technique described in Section 4 of Clark (1977) overcomes this issue.

R. M. Clark, Non-Parametric Estimation of a Smooth Regression Function, Journal of the Royal Statistical Society. Series B (Methodological), 1977, Vol. 39, No. 1 (1977), pp. 107-113 https://www.jstor.org/stable/2984885

---

The documentation for this class was generated from the following files:

• SmoothingSpline.h
• CrossValidatedSmoothing.C