Fitting Smoothing Spline ANOVA Models with Correlated Data

Description

Fit smoothing spline ANOVA models with correlated Gaussian data. The symbolic model specification via formula follows the same rules as in lm.

Usage

ssanova9(formula, type=NULL, data=list(), subset, offset,
         na.action=na.omit, partial=NULL, method="v", alpha=1.4,
         varht=1, id.basis=NULL, nbasis=NULL, seed=NULL, cov,
         skip.iter=FALSE)

para.arma(fit)

Arguments

Details

The model specification via formula is intuitive. For example, y~x1*x2 yields a model of the form

y = C + f_{1}(x1) + f_{2}(x2) + f_{12}(x1,x2) + e

with the terms denoted by "1", "x1", "x2", and "x1:x2".

The model terms are sums of unpenalized and penalized terms. Attached to every penalized term there is a smoothing parameter, and the model complexity is largely determined by the number of smoothing parameters.

A subset of the observations are selected as "knots." Unless specified via id.basis or nbasis, the number of "knots" q is determined by max(30,10n^{2/9}), which is appropriate for the default cubic splines for numerical vectors.

Using q "knots," ssanova calculates an approximate solution to the penalized least squares problem using algorithms of the order O(nq^{2}), which for q<<n scale better than the O(n^{3}) algorithms of ssanova0. For the exact solution, one may set q=n in ssanova, but ssanova0 would be much faster.

Value

ssanova9 returns a list object of class c("ssanova9","ssanova").

The method summary.ssanova9 can be used to obtain summaries of the fits. The method predict.ssanova can be used to evaluate the fits at arbitrary points along with standard errors. The method project.ssanova9 can be used to calculate the Kullback-Leibler projection for model selection. The methods residuals.ssanova and fitted.ssanova extract the respective traits from the fits.

para.arma returns the fitted ARMA coefficients for cov=list("arma",c(p,q)) in the call to ssanova9.

Skipping Theta Iteration

For the selection of multiple smoothing parameters, nlm is used to minimize the selection criterion such as the GCV score. When the number of smoothing parameters is large, the process can be time-consuming due to the great amount of function evaluations involved.

The starting values for the nlm iteration are obtained using Algorith 3.2 in Gu and Wahba (1991). These starting values usually yield good estimates themselves, leaving the subsequent quasi-Newton iteration to pick up the "last 10%" performance with extra effort many times of the initial one. Thus, it is often a good idea to skip the iteration by specifying skip.iter=TRUE, especially in high-dimensions and/or with multi-way interactions.

skip.iter=TRUE could be made the default in future releases.

Note

The results may vary from run to run. For consistency, specify id.basis or set seed.

Author(s)

Chong Gu, chong@stat.purdue.edu

References

Han, C. and Gu, C. (2008), Optimal smoothing with correlated data, Sankhya, 70-A, 38–72.

Examples

x <- runif(100); y <- 5 + 3*sin(2*pi*x) + rnorm(x)
## independent fit
fit <- ssanova9(y~x,cov=list("known",diag(1,100)))
## AR(1) fit
fit <- ssanova9(y~x,cov=list("arma",c(1,0)))
para.arma(fit)
## MA(1) fit
e <- rnorm(101); e <- e[-1]-.5*e[-101]
x <- runif(100); y <- 5 + 3*sin(2*pi*x) + e
fit <- ssanova9(y~x,cov=list("arma",c(0,1)))
para.arma(fit)
## Clean up
## Not run: rm(x,y,e,fit)