R/fosr.perm.R

In refunders/refund: Regression with Functional Data

##' Permutation testing for function-on-scalar regression
##'
##' \code{fosr.perm()} is a wrapper function calling \code{fosr.perm.fit()},
##' which fits models to permuted data, followed by \code{fosr.perm.test()},
##' which performs the actual simultaneous hypothesis test.  Calling the latter
##' two functions separately may be useful for performing tests at different
##' significance levels.  By default, \code{fosr.perm()} produces a plot using
##' the plot function for class \code{fosr.perm}.
##'
##'
##' @aliases fosr.perm fosr.perm.fit fosr.perm.test plot.fosr.perm
##' @rdname fosr.perm
##' @param Y,fdobj the functional responses, given as either an \eqn{n\times d}
##' matrix \code{Y} or a functional data object (class \code{"\link[fda]{fd}"})
##' as in the \pkg{fda} package.
##' @param X the design matrix, whose columns represent scalar predictors.
##' @param con a row vector or matrix of linear contrasts of the coefficient
##' functions, to be restricted to equal zero.
##' @param X0 design matrix for the null-hypothesis model.  If \code{NULL}, the
##' null hypothesis is the intercept-only model.
##' @param con0 linear constraints for the null-hypothesis model.
##' @param argvals the \eqn{d} argument values at which the coefficient
##' functions will be evaluated.
##' @param lambda smoothing parameter value.  If \code{NULL}, the smoothing
##' parameter(s) will be estimated.  See \code{\link{fosr}} for details.
##' @param lambda0 smoothing parameter for null-hypothesis model.
##' @param multi.sp a logical value indicating whether separate smoothing
##' parameters should be estimated for each coefficient function.  Currently
##' must be \code{FALSE} if \code{method = "OLS"}.
##' @param nperm number of permutations.
##' @param prelim number of preliminary permutations.  The smoothing parameter
##' in the main permutations will be fixed to the median value from these
##' preliminary permutations.  If \code{prelim=0}, this is not done. Preliminary
##' permutations are not available when \code{multi.sp = TRUE} (hence the complicated default).
##' @param level significance level for the simultaneous test.
##' @param plot logical value indicating whether to plot the real- and
##' permuted-data pointwise F-type statistics.
##' @param xlabel x-axis label for plots.
##' @param title title for plot.
##' @param x object of class \code{fosr.perm}, outputted by \code{fosr.perm},
##' \code{fosr.perm.fit}, or \code{fosr.perm.test}.
##' @param \dots for \code{fosr.perm} and \code{fosr.perm.fit}, additional
##' arguments passed to \code{\link{fosr}}.  These arguments may include
##' \code{max.iter}, \code{method}, \code{gam.method}, and \code{scale}.  For
##' \code{plot.fosr.perm}, graphical parameters (see \code{\link{par}}) for the
##' plot.
##' @return \code{fosr.perm} or \code{fosr.perm.test} produces an object of
##' class \code{fosr.perm}, which is a list with the elements below.
##' \code{fosr.perm.fit} also outputs an object of this class, but without the
##' last five elements. \item{F}{pointwise F-type statistics at each of the
##' points given by \code{argvals}.} \item{F.perm}{a matrix, each of whose rows
##' gives the pointwise F-type statistics for a permuted data set.}
##' \item{argvals}{points at which F-type statistics are computed.}
##' \item{lambda.real}{smoothing parameter(s) for the real-data fit.}
##' \item{lambda.prelim}{smoothing parameter(s) for preliminary permuted-data
##' fits.} \item{lambda.perm}{smoothing parameter(s) for main permuted-data
##' fits.} \item{lambda0.real, lambda0.prelim, lambda0.perm}{as above, but for
##' null hypothesis models.} \item{level}{significance level of the test.}
##' \item{critval}{critical value for the test.} \item{signif}{vector of
##' logical values indicating whether significance is attained at each of the
##' points \code{argvals}.} \item{n2s}{subset of 1, ...,
##' length(argvals) identifying the points at which the test statistic
##' changes from non-significant to significant.} \item{s2n}{points at which
##' the test statistic changes from significant to non-significant.}
##' @author Philip Reiss \email{phil.reiss@@nyumc.org} and Lan Huo
##' @seealso \code{\link{fosr}}
##' @references Reiss, P. T., Huang, L., and Mennes, M. (2010).  Fast
##' function-on-scalar regression with penalized basis expansions.
##' \emph{International Journal of Biostatistics}, 6(1), article 28.  Available
##' at \url{https://pubmed.ncbi.nlm.nih.gov/21969982/}
##' @examples
##'
##' \dontrun{
##' # Test effect of region on mean temperature in the Canadian weather data
##' # The next two lines are taken from the fRegress.CV help file (package fda)
##' smallbasis  <- create.fourier.basis(c(0, 365), 25)
##' tempfd <- smooth.basis(day.5,
##'           CanadianWeather$dailyAv[,,"Temperature.C"], smallbasis)$fd
##'
##' Xreg = cbind(1, model.matrix(~factor(CanadianWeather$region)-1))
##' conreg = matrix(c(0,1,1,1,1), 1)   # constrain region effects to sum to 0
##'
##' # This is for illustration only; for a real test, must increase nperm
##' # (and probably prelim as well)
##' regionperm = fosr.perm(fdobj=tempfd, X=Xreg, con=conreg, method="OLS", nperm=10, prelim=3)
##'
##' # Redo the plot, using axisIntervals() from the fda package
##' plot(regionperm, axes=FALSE, xlab="")
##' box()
##' axis(2)
##' axisIntervals(1)
##' }
##'
##' @export
fosr.perm <-
function(Y=NULL, fdobj=NULL, X, con=NULL, X0=NULL, con0=NULL, argvals = NULL,
lambda=NULL, lambda0=NULL, multi.sp=FALSE, nperm, level=.05, plot=TRUE, xlabel="", title=NULL, prelim=if (multi.sp) 0 else 15, ...) {
    fpobj1 = fosr.perm.fit(Y=Y, fdobj=fdobj, X=X, con=con, X0=X0, con0=con0, argvals=argvals, lambda=lambda, lambda0=lambda0, multi.sp=multi.sp, nperm=nperm, prelim=prelim, ...)
    fpobj2 = fosr.perm.test(fpobj1, level=level)
    if (plot) plot(fpobj2, xlabel=xlabel, title=title)
    fpobj2
}