Dear,
When i multiply the y of a regression by 10, I would expect that the
coefficient would be multiply by 10 and the z value to stay constant. Here
some reproducible code to support the case.
*Ex 1*
library(mvtnorm)
library(VGAM)
set.seed(1)
x=rmvnorm(1000,sigma=matrix(c(1,0.75,0.75,1),2,2))
summary(vglm("y~x",family=studentt2,data=data.frame("y"=x[,1],"x"=x[,2])))
summary(vglm("y~x",family=studentt2,data=data.frame("y"=x[,1]*10,"x"=x[,2])))
summary(vglm("y~x",family=cauchy1,data=data.frame("y"=x[,1],"x"=x[,2])))
summary(vglm("y~x",family=cauchy1,data=data.frame("y"=x[,1]*10,"x"=x[,2])))
summary(vglm("y~x",family=hypersecant,data=data.frame("y"=x[,1],"x"=x[,2])))
summary(vglm("y~x",family=hypersecant,data=data.frame("y"=x[,1]*10,"x"=x[,2])))
*Ex 2*
library(VGAM)
tdata <- data.frame(x2 = runif(nn <- 1000))
tdata <- transform(tdata, y1 = rt(nn, df = exp(exp(0.5 - x2))),
y2 = rt(nn, df = exp(exp(0.5 - x2))))
fit1 <- vglm(y1 ~ x2, studentt, tdata, trace = TRUE)
tdata$y1=tdata$y1*100
fit2 <- vglm(y1 ~ x2, studentt, tdata, trace = TRUE)
coef(fit1, matrix = TRUE)
coef(fit2, matrix = TRUE)
I also feel that often VGAM package (vglm function) dont converge and just
stops. Do you know a reliable package with a lot of available distribution ?
Thanks,
Olivier
[[alternative HTML version deleted]]
Dear, I dont see my topic in the daily summary : Thanks ---------- Forwarded message ---------- From: Olivier Merle <oliviermerle35@gmail.com> Date: 2013/4/17 Subject: Bug in VGAM z value and coefficient ? To: r-help@r-project.org Dear, When i multiply the y of a regression by 10, I would expect that the coefficient would be multiply by 10 and the z value to stay constant. Here some reproducible code to support the case. *Ex 1* library(mvtnorm) library(VGAM) set.seed(1) x=rmvnorm(1000,sigma=matrix(c(1,0.75,0.75,1),2,2)) summary(vglm("y~x",family=studentt2,data=data.frame("y"=x[,1],"x"=x[,2]))) summary(vglm("y~x",family=studentt2,data=data.frame("y"=x[,1]*10,"x"=x[,2]))) summary(vglm("y~x",family=cauchy1,data=data.frame("y"=x[,1],"x"=x[,2]))) summary(vglm("y~x",family=cauchy1,data=data.frame("y"=x[,1]*10,"x"=x[,2]))) summary(vglm("y~x",family=hypersecant,data=data.frame("y"=x[,1],"x"=x[,2]))) summary(vglm("y~x",family=hypersecant,data=data.frame("y"=x[,1]*10,"x"=x[,2]))) *Ex 2* library(VGAM) tdata <- data.frame(x2 = runif(nn <- 1000)) tdata <- transform(tdata, y1 = rt(nn, df = exp(exp(0.5 - x2))), y2 = rt(nn, df = exp(exp(0.5 - x2)))) fit1 <- vglm(y1 ~ x2, studentt, tdata, trace = TRUE) tdata$y1=tdata$y1*100 fit2 <- vglm(y1 ~ x2, studentt, tdata, trace = TRUE) coef(fit1, matrix = TRUE) coef(fit2, matrix = TRUE) I also feel that often VGAM package (vglm function) dont converge and just stops. Do you know a reliable package with a lot of available distribution ? Thanks, Olivier [[alternative HTML version deleted]]