R/predict.ssanova.R

In gss: General Smoothing Splines

## Calculate prediction and Bayesian SE from ssanova objects
predict.ssanova <- function(object,newdata,se.fit=FALSE,
                            include=c(object$terms$labels,object$lab.p),...)
{
    nnew <- nrow(newdata)
    nbasis <- length(object$id.basis)
    nnull <- length(object$d)
    nz <- length(object$b)
    nn <- nbasis + nnull + nz
    labels.p <- object$lab.p
    ## Extract included terms
    term <- object$terms
    philist <- rklist <- NULL
    s <- r <- NULL
    nq <- 0
    for (label in include) {
        if (label=="1") {
            philist <- c(philist,term[[label]]$iphi)
            s <- cbind(s,rep(1,len=nnew))
            next
        }
        if (label%in%labels.p) next
        if (label=="offset") next
        xnew <- newdata[,term[[label]]$vlist]
        x <- object$mf[object$id.basis,term[[label]]$vlist]
        nphi <- term[[label]]$nphi
        nrk <- term[[label]]$nrk
        if (nphi) {
            iphi <- term[[label]]$iphi
            phi <- term[[label]]$phi
            for (i in 1:nphi) {
                philist <- c(philist,iphi+(i-1))
                s <- cbind(s,phi$fun(xnew,nu=i,env=phi$env))
            }
        }
        if (nrk) {
            irk <- term[[label]]$irk
            rk <- term[[label]]$rk
            for (i in 1:nrk) {
                rklist <- c(rklist,irk+(i-1))
                nq <- nq+1
                r <- array(c(r,rk$fun(xnew,x,nu=i,env=rk$env,out=TRUE)),c(nnew,nbasis,nq))
            }
        }
    }
    if (!is.null(object$partial)) {
        vars.p <- as.character(attr(object$partial$mt,"variables"))[-1]
        facs.p <- attr(object$partial$mt,"factors")
        vlist <- vars.p[as.logical(apply(facs.p,1,sum))]
        for (lab in labels.p) {
            if (lab%in%include) {
                vlist.wk <- vars.p[as.logical(facs.p[,lab])]
                vlist <- vlist[!(vlist%in%vlist.wk)]
            }
        }
        if (length(vlist)) {
            for (lab in vlist) newdata[[lab]] <- 0
        }
        matx.p <- model.matrix(object$partial$mt,newdata)[,-1,drop=FALSE]
        matx.p <- sweep(matx.p,2,object$partial$center)
        matx.p <- sweep(matx.p,2,object$partial$scale,"/")
        nu <- nnull-dim(matx.p)[2]
        for (label in labels.p) {
            nu <- nu+1
            if (label%in%include) {
                philist <- c(philist,nu)
                s <- cbind(s,matx.p[,label])
            }
        }
    }
    r.wk <- matrix(0,nnew,nbasis)
    nq <- 0
    for (i in rklist) {
        nq <- nq + 1
        r.wk <- r.wk + 10^object$theta[i]*r[,,nq]
    }
    ## random effects
    if (nz) {
        if (is.null(newdata$random)) z.wk <- matrix(0,nnew,nz)
        else z.wk <- newdata$random
        r.wk <- cbind(r.wk,z.wk)
    }
    ## Compute posterior mean
    nphi <- length(philist)
    pmean <- as.vector(r.wk%*%c(object$c,object$b))
    if (nphi) pmean <- pmean + as.vector(s%*%object$d[philist])
    if (any(include=="offset")) {
        if (is.null(model.offset(object$mf)))
            stop("gss error: no offset in the fit")
        offset <- newdata$offset
        if (is.null(offset)) offset <- newdata$"(offset)"
        if (is.null(offset)) stop("gss error: missing offset")
        pmean <- pmean + offset
    }
    if (se.fit) {
        b <- object$varht/10^object$nlambda
        ## Compute posterior variance
        ss <- matrix(0,nnull,nnew)
        if (!is.null(philist)) ss[philist,] <- t(s)
        rr <- t(r.wk%*%object$se.aux$vec)
        wk <- object$se.aux$hfac%*%rbind(ss,rr)
        pse <- sqrt(b*apply(wk^2,2,sum))
        list(fit=pmean,se.fit=pse)
    }
    else pmean
}