R/predict.mactivate_fit_gradient_logistic_01.R

In mactivate: Multiplicative Activation

predict.mactivate_fit_gradient_logistic_01 <-
function (object, X0, U0 = NULL, mcols, ...) 
{
    if (is.null(U0)) {
        U0 <- X0
    }
    xmcol <- mcols
    mobj_use <- object[[xmcol + 1]]
    What_use <- mobj_use[["What"]]
    X0star <- f_mactivate(U = U0, W = What_use)
    Xfull <- cbind(rep(1, nrow(X0)), X0, X0star)
    bc_hat <- c(mobj_use[["b0hat"]], mobj_use[["bbhat"]], mobj_use[["cchat"]])
    y0hat <- Xfull %*% bc_hat
    p0hat <- 1/(exp(-y0hat) + 1)
    ls_out <- list(y0hat = y0hat, p0hat = p0hat)
    return(ls_out)
}