funeigen {funreg}

Perform eigenfunction decomposition on functional covariate

Description

A function to do the eigenfunction decomposition as part of a penalized functional regression as in Goldsmith et al. (2011)

Usage

funeigen(id, time, x, num.bins = 35, preferred.num.eigenfunctions = 30)

Arguments

id

A vector of subject ID's.

time

A vector of measurement times.

x

A single functional predictor represented as a vector or a one-column matrix.

num.bins

The number of knots used in the spline basis for the beta function. The default is based on the Goldsmith et al. (2011) sample code.

preferred.num.eigenfunctions

The number of eigenfunctions to use in approximating the covariance function of x (see Goldsmith et al., 2011)

Note

The algorithm for this function follows that of "sparse_simulation.R", which was written on Nov. 13, 2009, by Jeff Goldsmith; Goldsmith noted that he used some code from Chongzhi Di for the part about handling sparsity. "sparse_simulation.R" was part of the supplementary material for Goldsmith, Bobb, Crainiceanu, Caffo, and Reich (2011). The num.bins parameter corresponds to N.fit in Goldsmith et al, sparse_simulation.R and preferred.num.eigenfunctions corresponds to Kz in Goldsmith et al.

References

Goldsmith, J., Bobb, J., Crainiceanu, C. M., Caffo, B., and Reich, D. (2011). Penalized functional regression. Journal of Computational and Graphical Statistics, 20(4), 830-851. DOI: 10.1198/jcgs.2010.10007.

See Also

fitted.funeigen, link{plot.funeigen}