Predict responses and latent scores from PLS fits

These prediction helpers reconstruct the response matrix and latent component scores for partial least squares (PLS) models fitted inside the Cox-PLS toolbox. They support group PLS, sparse PLS, sparse-group PLS, and classical PLS models created by sgPLS::gPLS(), sgPLS::sPLS(), sgPLS::sgPLS(), or plsRcox::pls.cox().

Usage

# S3 method for class 'gPLS'
predict(object, newdata, scale.X = TRUE, scale.Y = TRUE, ...)

# S3 method for class 'pls.cox'
predict(object, newdata, scale.X = TRUE, scale.Y = TRUE, ...)

# S3 method for class 'sPLS'
predict(object, newdata, scale.X = TRUE, scale.Y = TRUE, ...)

# S3 method for class 'sgPLS'
predict(object, newdata, scale.X = TRUE, scale.Y = TRUE, ...)

Arguments

object

A fitted PLS model returned by sgPLS::gPLS(), sgPLS::sPLS(), sgPLS::sgPLS(), or plsRcox::pls.cox().

newdata

Numeric matrix or data frame with the same number of columns as the training design matrix used when fitting object.

scale.X, scale.Y

Logical flags indicating whether the predictors and responses supplied in newdata should be centred and scaled according to the training statistics stored in object.

...

Unused arguments included for compatibility with the generic stats::predict() signature.

Value

A list containing reconstructed responses, latent component scores, and regression coefficients. The exact elements depend on the specific PLS algorithm but always include components named predict, variates, and B.hat.

References

Bastien, P., Bertrand, F., Meyer, N., & Maumy-Bertrand, M. (2015). Deviance residuals-based sparse PLS and sparse kernel PLS for censored data. Bioinformatics, 31(3), 397–404. doi:10.1093/bioinformatics/btu660

See also

coxgpls(), coxsgpls(), coxspls_sgpls(), and coxDKgplsDR() for Cox model wrappers that return PLS fits using these prediction methods.

Examples

n <- 100
sigma.gamma <- 1
sigma.e <- 1.5
p <- 400
q <- 500
theta.x1 <- c(rep(1, 15), rep(0, 5), rep(-1, 15), rep(0, 5), rep(1.5,15), 
              rep(0, 5), rep(-1.5, 15), rep(0, 325))
theta.x2 <- c(rep(0, 320), rep(1, 15), rep(0, 5), rep(-1, 15), rep(0, 5),
              rep(1.5, 15), rep(0, 5), rep(-1.5, 15), rep(0, 5))
theta.y1 <- 1
theta.y2 <- 1

Sigmax <- matrix(0, nrow = p, ncol = p)
diag(Sigmax) <- sigma.e ^ 2
Sigmay <- matrix(0,nrow = 1, ncol = 1)
diag(Sigmay) <- sigma.e ^ 2

set.seed(125)

gam1 <- rnorm(n)
gam2 <- rnorm(n)

X <- matrix(c(gam1, gam2), ncol = 2, byrow = FALSE) %*% matrix(c(theta.x1, theta.x2),
 nrow = 2, byrow = TRUE) + mvtnorm::rmvnorm(n, mean = rep(0, p), sigma =
 Sigmax, method = "svd")
Y <- matrix(c(gam1, gam2), ncol = 2, byrow = FALSE) %*% matrix(c(theta.y1, theta.y2), 
nrow = 2, byrow = TRUE) + rnorm(n,0,sd=sigma.e)

ind.block.x <- seq(20, 380, 20)

model.gPLS <- sgPLS::gPLS(X, Y, ncomp = 2, mode = "regression", keepX = c(4, 4), 
                   keepY = c(4, 4), ind.block.x = ind.block.x)
head(predict(model.gPLS, newdata = X)$variates)
#>            dim 1      dim 2
#> [1,]   1.7433859 -1.0107806
#> [2,]   0.3536207  5.7590437
#> [3,]   4.6851991 -1.7566289
#> [4,]   0.2587842 -0.6634727
#> [5,]   1.5699343  0.4017589
#> [6,] -10.5958505 -2.3180566