R/olse.r

In lrmest: Different Types of Estimators to Deal with Multicollinearity

ols<-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)
    y <- model.response(cal)
    md <- attr(cal, "terms")
    x <- model.matrix(md, cal, contrasts)
    s <- t(x) %*% x
    xin <- solve(s)
    bb <- xin %*% t(x) %*% y
    colnames(bb) <- c("Estimate")
    ev <- (t(y) %*% y - t(bb) %*% t(x) %*% y)/(NROW(x) - NCOL(x))
    ev <- diag(ev)
    dbb <- ev * xin
    Standard_error <- sqrt(diag(abs(dbb)))
    dbt <- t(bb)
    sdbd_inv <- (sqrt(diag(abs(dbb))))^-1
    sdbd_inv_mat <- diag(sdbd_inv)
    if (NCOL(dbt) == 1L) 
        tbb <- dbt * sdbd_inv
    else tbb <- dbt %*% sdbd_inv_mat
    tst <- t(tbb)
    pval <- t(2 * pt(-abs(tbb), df <- (NROW(x) - NCOL(x))))
    colnames(pval) <- c("p_value")
    colnames(tst) <- c("t_statistic")
    mse1 <- sum(diag(dbb))
    names(mse1) <- c("MSE")
    mse1 <- round(mse1, digits <- 4L)
    if (NCOL(x) == 1L) 
        svec <- x
    else svec <- x[, -1L]
    smat <- as.matrix(svec)
    invcor <- solve(cor(smat))
    avecor <- sum(diag(invcor))/NCOL(x[, -1L])
    con <- if (avecor > 10L) 
        warning("There is a multicollinearity")
    else "There is not multicollinearity"
    ans <- cbind(bb, Standard_error, tst, pval)
    ans1 <- round(ans, digits <- 4L)
    adw <- list(`*****Ordinary Least Square Estimator******` = ans1, 
        `*****Mean square error value*****` = mse1)
    adw
}