R/wxgb.error.R

Defines functions Loss evalerror_reg eval_loss evalerror_bin

Documented in evalerror_bin evalerror_reg eval_loss

#' Weighted mean error of binary outcome for complex survey data
#'
#' @description
#' A function to evaluate XGBoost models by weighted mean error for binary outcome (`0/1`).  This function is a customized error function to
#' plug in the function \code{wXGBoost()} for complex survey data.
#'
#' @param preds Predicted binary values (`0/1`) from the model.
#' @param dtrain An object of \code{xgboost::xgb.DMatrix} including `label`(response variable or binary outcome), `data`, and `weight`.
#' @return A list of two components:
#' - `metric` : A string labelled as `weighted.error`.
#' - `value`  : A numeric value of weighted mean error.
#'
#' @seealso [xgboost::xgb.train()] for relevant arguments and return values in detail.
#' @seealso [wXGBoost()] for detailed information on implementation.
#' @seealso [optim_wxgb_para()] for detailed optimization and examples.
#'
#
#'
#'@export
evalerror_bin<-function(preds,dtrain){
  labels<-getinfo(dtrain,"label")
  weights <- getinfo(dtrain,"weight")
  err <- stats::weighted.mean(labels!=(preds>0),weights,na.rm=TRUE)

  return(list(metric="weighted.mean.error", value=err)) }


#' Weighted mean of log loss (binary cross-entropy loss) for complex survey data
#'
#' @description
#' A function to evaluate XGBoost models by weighted mean log loss for logistic regression. This function estimates predicted
#' probabilities from the predicted binary outcomes and computes the weighted mean of log likelihood values for Bernoulli distribution.
#' This function is a customized error function for complex survey data.
#'
#' @param preds Predicted binary values (`0/1`) from the model.
#' @param dtrain An object of \code{xgb.DMatrix} including `label`(response variable or binary outcome), `data`,`weight`.
#' @return A list of two components:
#' - `metric`: A string labelled as `weighted.mean.logloss`.
#' - `value` : A numeric value of weighted mean of log loss.
#' @seealso [xgboost::xgb.train()] for relevant arguments and return values in detail.
#' @seealso [wXGBoost()] for detailed information on implementation.
#' @seealso [optim_wxgb_para()] for detailed optimization and examples.
#'
#'
#' @export
eval_loss<- function(preds,dtrain){
  labels<-getinfo(dtrain,"label")
  weights <- getinfo(dtrain,"weight")
  preds <- 1/(1+exp(-preds))
  loss<- Loss(preds,labels,"binary:logistic")
  wloss <- stats::weighted.mean(loss,weights,na.rm=TRUE)
  # wloss<- sum(loss*weights)/sum(weights)
  return(list(metric="weighted.mean.logloss",value=wloss))
}


#' Weighted mean squared error for complex survey data
#'
#' @description
#'  A function to evaluate XGBoost models by weighted mean squared error for linear regression.
#'  This function is a customized error function for complex survey data.
#'
#' @param preds Predicted values from the model
#' @param dtrain An object of \code{xgboost::xgb.DMatrix} including `label`(response variable), `data`, and `weight`.
#' @return A list of two components:
#'- `metric`: A string labelled as `weighted.mse`.
#' - `value` : A numeric value of weighted mean squared error (MSE).
#' @seealso [xgboost::xgb.train()] for relevant arguments and return values in detail.
#' @seealso [wXGBoost()] for detailed information on implementation.
#' @seealso [optim_wxgb_para()] for detailed optimization and examples.
#'
#'
#'
#' @export
evalerror_reg <- function(preds,dtrain){
  labels <- getinfo(dtrain,"label")
  weights<- getinfo(dtrain,"weight")
  err<- stats::weighted.mean((labels-preds)^2,weights,na.rm=TRUE)

  return(list(metric="weighted.mse", value=err))

}

Loss <- function(y.est, y, objective = c("reg:squarederror","binary:logistic")){

  if(objective == "reg:squarederror"){
    l <- (y.est - y)^2
  }

  if(objective == "binary:logistic"){


    l <- rep(NA, length(y))
    l[which(y==1)] <- -log(y.est[which(y==1)])
    l[which(y==0)] <- -log(1-y.est[which(y==0)])

  }

  return(l)

}

Try the MLSurvey package in your browser

Any scripts or data that you put into this service are public.

MLSurvey documentation built on Nov. 18, 2026, 5:11 p.m.