In cugliari/bagdenest: Bagging of Density Estimators

knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>",
  fig.path = "man/figures/README-",
  out.width = "100%"
)

bagdenest

The goal of bagdenest is to provide density estimation by averaging classical density estimators such as the histogram, the frequency polygon and the kernel density estimators obtained over different bootstrap samples of the original data.

License

This package is free and open source software, licensed under GPL-3.

Installation

You can install (comming soon) the released version of bagdenest from CRAN with:

install.packages("bagdenest")

And the development version from GitHub with:

# install.packages("devtools")
devtools::install_github("cugliari/bagdenest")

Example

This is a basic example which shows you how to display densities estimator errors using bagdenest:

library(bagdenest)

simulations=function(n = 100, M = 10, B = 150){
  vec = c(1,3,5,7,8,11,10,13,17,19,20,21)
  AA = matrix(0, nrow = length(vec), ncol = 7)
  colnames(AA)=c("H","FP","Kde", "BagHist", "BagFP", "BagKde", "Rash")
  rownames(AA)=c("normal","chi2_10","MIX_1","MIx_2","CLAW","Triangular",
                 "DENS2_rigollet","uniforme","SmooComb","DistBim","DENS1_rigollet","MIX_Unif")
  for(i in 1:M){
    A = matrix(NA, nrow = length(vec), ncol = 7)
    for(ll in 1:length(vec)){
      dd = gendata(vec[ll],n)
      bopt = bropt(dd$train)$opt  
      zz = hist(dd$train,breaks=mybreaks(dd$train,nbr=bopt),plot=F)
      bhist = BagHistfp(xx=dd$train, grid=dd$test, B)
      modelrash = rash(dd$train, grid = dd$test, nbr = bopt, B)
      modelbagkde <- Bagkde(xx = dd$train, grid = dd$test, B)
      A[ll,]=c(error(dd$dobs,predict_hist(zz,dd$test))[1],
               error(dd$dobs,approxfun(x=zz$mids,y=zz$density)(dd$test))[1],
               error(dd$dobs,onekdeucv(dd$train,dd$test))[1],
               error(dd$dobs,bhist$bh)[1],
               error(dd$dobs,bhist$bhfp)[1],
               error(dd$dobs,modelbagkde)[1],
               error(dd$dobs,modelrash)[1])
      }
    AA=AA+A
  }
  AA=AA/M
  print(AA)
}
simulations(n = 100, M = 2, B = 15)