R/checkmt.r

In lrmest: Different Types of Estimators to Deal with Multicollinearity

checkm<-function (formula, data, na.action, ...) 
{
    cal <- match.call(expand.dots = FALSE)
    mat <- match(c("formula", "data", "na.action"), names(cal))
    cal <- cal[c(1L, mat)]
    cal[[1L]] <- as.name("model.frame")
    cal <- eval(cal)
    md <- attr(cal, "terms")
    dmat <- model.matrix(md, cal, contrasts)
    if (NCOL(dmat) == 1L) 
        svec <- dmat
    else svec <- dmat[, -1L]
    smat <- as.matrix(svec)
    cor_mat <- cor(smat)
    eimat <- eigen(cor_mat)
    cn <- sqrt(abs(max(eimat$values)/min(eimat$values)))
    invcor <- solve(cor(smat))
    avecor <- t(diag(invcor))
    res <- list(`*****Condition number*****` = cn, `*****VIF*****` = avecor)
    res
}