R/summary_rewlr.R

In zaenalium/rewlr: Rare Event Weighted Logistics Regression

#' Summarizing the rare event weighted logistic regreression
#'
#' @description summary method for class "rewlr".
#' @param model an object of class "rewlr"
#' @param digit the number of significant digits to use when printing
#' @return rewlr returns output like glm, use function summary() to obtain the summary coefficients and others. The detail are shown in the following list:
#' \itemize{
#' \item coefficients - a named vector of coefficients.
#' \item fitted.values - return the prediction using the training data resulting probablity.
#' \item deviance - up to a constant, minus twice the maximized log-likelihood. Where sensible, the constant is chosen so that a saturated model has deviance zero.
#' \item AIC - A version of Akaike's An Information Criterion, minus twice the maximized log-likelihood plus twice the number of parameter.
#' \item null.deviance - The deviance for the null model, comparable with deviance. The null model will include the offset, and an intercept if there is one in the model.
#' \item df.residual - the residual degrees of freedom.
#' \item df.null - the residual degrees of freedom for the null model.
#' \item auc - an area under ROC curve
#' }
#' @examples library(rewlr)
#' data <- iris[1:55, ]
#' data$Species <- ifelse(data$Species == "setosa",0,1)
#' #Supposed that current sample data has 9 percent of rare event data, and the population has 2 percent of those rare event data.
#' (weight0 = (1 - 0.09)/(1-0.02))
#' (weight1 = (0.09)/(0.02))
#' iter = 1000; tol = 0.00001
#'
#' fit <- rewlr(Species, data = data, weight0 = weight0, weight1 = weight1)
#' summary(fit)
#' p <- predict(fit, newdata = iris[c(1:22,61) , ])
#' @export summary.rewlr
summary.rewlr <- function(model, digits = max(3L, getOption("digits") - 3L), ...) {
  dn <- c("Estimate", "Std. Error")
    pvalue <- 2 * pnorm(-abs(model$wald))
    coef.table <- cbind(model$B, model$std_error,exp(model$B), model$wald, pvalue)
    dimnames(coef.table) <- list(rownames(model$B), c(dn, "OddRatio",
                                                  "Wald", "Pr(>|z|)"))
    cat("\nCoefficients: \n")
    printCoefmat(coef.table, digits = digits)

    cat("\nDegrees of Freedom:", model$df_null, "Total (i.e. Null); ",
        model$df_res, "Residual\n")
    #if (nzchar(mess <- naprint(x$na.action)))
    #  cat("  (", mess, ")\n", sep = "")
    cat("Null Deviance:\t   ", format(signif(model$null_dev,
                                             digits)), "\nResidual Deviance:", format(signif(model$res_dev,
                                                                                             digits)), "\tAIC:", format(signif(model$aic, digits)))
    cat("\n")
    cat(#"Area under the curve:\t", format(signif(model$auc,digits)),
      "\tMcFadden R2 :\t   ", format(signif(model$PseudoR2,digits)),
        "\nOveral Wald Test:", format(signif(model$wald_all,digits)),
        "\tp-value = ", format(signif(2 * pchisq(-abs(model$wald_all), nrow(coef.table)), 3 )))
    invisible(model)
}