[R] Zero errors : Bug in my R code ?

varin sacha v@rin@@ch@ @ending from y@hoo@fr
Wed Apr 25 23:24:14 CEST 2018


Dear R-experts,

I guess I have a problem with my fast function (fast tau estimator) here below. Indeed, zero errors look highly suspicious. I guess there is a bug in my R code. How could I correct my R code ?

# install.packages( "robustbase" )
# install.packages( "MASS" )
# install.packages( "quantreg" )
# install.packages( "RobPer" )
# install.packages("devtools")  # install_github("kloke/hbrfit") #install.packages('http://www.stat.wmich.edu/mckean/Stat666/Pkgs/npsmReg2_0.1.1.tar.gz') 
#install.packages("DMwR")

library(robustbase)
library(MASS)
library(quantreg)
library(RobPer)
library(hbrfit)
library("devtools") 
library("DMwR")

bmi=c(23,43,12,1,23,4,3,4,5,5,6,5,5,34,67,87,32,12,23,19) 
glucose=c(98,34,21,13,2,43,16,43,28,93,11,14,16,43,23,65,42,16,54,32)
crp=c(56,34,42,32,1,2,54,47,67,65,38,95,45,76,67,87,35,41,34,54)
newdata=data.frame(bmi,glucose,crp)
 
reg1 <- lm( crp ~ bmi+glucose,data=newdata)
DMwR::regr.eval(newdata$crp,reg1$fitted.values)

reg <- lmrob( crp ~ bmi+glucose,data=newdata)
DMwR::regr.eval(newdata$crp,reg$fitted.values)

Huber <- rlm( crp ~ bmi+glucose,data=newdata)
DMwR::regr.eval(newdata$crp,Huber$fitted.values)

Tukey <- rlm( crp ~ bmi+glucose, data=newdata, psi = psi.bisquare )
DMwR::regr.eval(newdata$crp,Tukey$fitted.values)

L1 <- rq( crp ~ bmi+glucose, data=newdata, tau = 0.5 )
DMwR::regr.eval(newdata$crp,L1$fitted.values)

fast <- FastTau(model.matrix(~newdata$bmi+newdata$glucose),newdata$crp)
DMwR::regr.eval(newdata$crp,fast$fitted.values)

HBR<-hbrfit(crp ~ bmi+glucose)
DMwR::regr.eval(newdata$crp,HBR$fitted.values)




More information about the R-help mailing list