Numerische Mathematik

, Volume 31, Issue 4, pp 377–403

Smoothing noisy data with spline functions

Estimating the correct degree of smoothing by the method of generalized cross-validation
• Peter Craven
• Grace Wahba
Article

Summary

Smoothing splines are well known to provide nice curves which smooth discrete, noisy data. We obtain a practical, effective method for estimating the optimum amount of smoothing from the data. Derivatives can be estimated from the data by differentiating the resulting (nearly) optimally smoothed spline.

We consider the modely i (t i )+ε i ,i=1, 2, ...,n,ti∈[0, 1], wheregW 2 (m) ={f:f,f′, ...,f(m−1) abs. cont.,f(m)∈ℒ2[0,1]}, and the {ε i } are random errors withEε i =0,Eε i ε j 2δ ij . The error variance σ2 may be unknown. As an estimate ofg we take the solutiongn, λ to the problem: Findf∈W 2 (m) to minimize$$\frac{1}{n}\sum\limits_{j = 1}^n {(f(t_j ) - y_j )^2 + \lambda \int\limits_0^1 {(f^{(m)} (u))^2 du} }$$. The functiongn, λ is a smoothing polynomial spline of degree 2m−1. The parameter λ controls the tradeoff between the “roughness” of the solution, as measured by$$\int\limits_0^1 {[f^{(m)} (u)]^2 du}$$, and the infidelity to the data as measured by$$\frac{1}{n}\sum\limits_{j = 1}^n {(f(t_j ) - y_j )^2 }$$, and so governs the average square errorR(λ; g)=R(λ) defined by
$$R(\lambda ) = \frac{1}{n}\sum\limits_{j = 1}^n {(g_{n,\lambda } (t_j ) - g(t_j ))^2 }$$
. We provide an estimate$$\hat \lambda$$, called the generalized cross-validation estimate, for the minimizer ofR(λ). The estimate$$\hat \lambda$$ is the minimizer ofV(λ) defined by$$V(\lambda ) = \frac{1}{n}\parallel (I - A(\lambda ))y\parallel ^2 /\left[ {\frac{1}{n}{\text{Trace(}}I - A(\lambda ))} \right]^2$$, wherey=(y1, ...,yn)t andA(λ) is then×n matrix satisfying(gn, λ (t1), ...,gn, λ (tn))t=A (λ) y. We prove that there exist a sequence of minimizers$$\tilde \lambda = \tilde \lambda (n)$$ ofEV(λ), such that as the (regular) mesh{ti} i=1 n becomes finer,$$\mathop {\lim }\limits_{n \to \infty } ER(\tilde \lambda )/\mathop {\min }\limits_\lambda ER(\lambda ) \downarrow 1$$. A Monte Carlo experiment with several smoothg's was tried withm=2,n=50 and several values of σ2, and typical values of$$R(\hat \lambda )/\mathop {\min }\limits_\lambda R(\lambda )$$ were found to be in the range 1.01–1.4. The derivativeg′ ofg can be estimated by$$g'_{n,\hat \lambda } (t)$$. In the Monte Carlo examples tried, the minimizer of$$R_D (\lambda ) = \frac{1}{n}\sum\limits_{j = 1}^n {(g'_{n,\lambda } (t_j ) - } g'(t_j ))$$ tended to be close to the minimizer ofR(λ), so that$$\hat \lambda$$ was also a good value of the smoothing parameter for estimating the derivative.

Subject Classifications

MOS:65D10 CR:5.17 MOS:65D25

Preview

Unable to display preview. Download preview PDF.

References

1. 1.
Abramowitz, M., Stegun, I.: Handbook of mathematical functions with formulas, graphs, and mathematical tables. U.S. Department of Commerce, National Bureau of Standards Applied Mathematics Series No.55, pp. 803–819, 1964Google Scholar
2. 2.
Aronszajn, N.: Theory of reproducing kernels. Trans. Amer. Math. Soc.68, 337–404 (1950)Google Scholar
3. 3.
Golomb, M.: Approximation by periodic spline interpolants on uniform meshes. J. Approximation Theory1, 26–65 (1968)Google Scholar
4. 4.
Golub, G., Heath, M., Wahba, G.: Generalized cross validation as a method for choosing a good ridge parameter, to appear, TechnometricsGoogle Scholar
5. 5.
Golub, G., Reinsch, C.: Singular value decomposition and least squares solutions. Numer. Math.14, 403–420 (1970)Google Scholar
6. 6.
Hudson, H.M.: Empirical Bayes estimation. Technical Report #58, Stanford University, Department of Statistics, Stanford, Cal., 1974Google Scholar
7. 7.
Kimeldorf, G., Wahba, G.: A correspondence between Bayesian estimation on stochastic processes and smoothing by splines. Ann. Inst. Statist. Math.41, 495–502 (1970)Google Scholar
8. 8.
Mallows, C.L.: Some comments onC p. Technometrics15, 661–675 (1973)Google Scholar
9. 9.
Reinsch, C.M.: Smoothing by spline functions. Numer. Math.10, 177–183 (1967)Google Scholar
10. 10.
Reinsch, C.M.: Smoothing by spline functions, II. Numer. Math.16, 451–454 (1971)Google Scholar
11. 11.
Schoenberg, I.J.: Spline functions and the problem of graduation. Proc. Nat. Acad. Sci. (USA)52, 947–950 (1964)Google Scholar
12. 12.
Wahba, G.: Convergence rates for certain approximate solutions to first kind integral equations. J. Approximation Theory7, 167–185 (1973)Google Scholar
13. 13.
Wahba, G.: Smoothing noisy data with spline functions. Numer. Math.24, 383–393 (1975)Google Scholar
14. 14.
Wahba, G.: Practical approximate solutions to linear operator equations when the data are noisy. SIAM J. Numer. Anal.14, 651–667 (1977)Google Scholar
15. 15.
Wahba, G., Wold, S.: A completely automatic French curve: Fitting spline functions by crossvalidation. Comm. Statist.4, 1–17 (1975)Google Scholar
16. 16.
Wahba, G., Wold, S.: Periodic splines for spectral density estimation: The use of cross-validation for determining the degree of smoothing. Comm. Statist.4, 125–141 (1975)Google Scholar
17. 17.
Wahba, G.: A survey of some smoothing problems and the method of generalized cross validation for solving them. University of Wisconsin-Madison, Statistics Dept., Technical Report #457. In: Proceedings of the Conference on Applications of Statistics, Dayton, Ohio (P.R. Krishnaiah, ed.) June 14–18, 1976Google Scholar
18. 18.
Wahba, G.: Improper priors, spline smoothing and the problem of guarding against model errors in regression. J. Roy. Statist. Soc., Ser. B. To appearGoogle Scholar

Authors and Affiliations

• Peter Craven
• 1
• Grace Wahba
• 2
1. 1.The Computer LaboratoryThe University of LiverpoolLiverpoolEngland
2. 2.Department of StatisticsUniversity of WisconsinMadisonUSA

Personalised recommendations

Citearticle 