# R/confint.expectreg.R In expectreg: Expectile and Quantile Regression

#### Documented in confint.expectreg

```confint.expectreg <-
function(object, parm=NULL, level=0.95,...)
{
if(is.null(object\$covmat))
stop("No covariance matrix calculated.")
res = list()

if(any(object\$design[,1] != 1))
center = FALSE
else
center = TRUE

if(is.null(parm))
for(i in 1:length(object\$asymmetries))
{
res[[i]] = matrix(NA,nrow=nrow(object\$design),ncol=2)
colnames(res[[i]]) = c(paste(eval((1-level)/2),"%"),paste(eval((1+level)/2),"%"))
for(j in 1:nrow(object\$design))
{
deviation = qnorm((1+level)/2) * sqrt(t(object\$design[j,]) %*% object\$covmat[[i]] %*% object\$design[j,])
res[[i]][j,] = c(fitted(object)[j,i] - deviation, fitted(object)[j,i] + deviation)
}
}
else
{
nb = NULL
for(i in 1:length(object\$coefficients))
nb = c(nb,nrow(object\$coefficients[[i]]))

PB = NULL
Koff = NULL

for(k in 1:length(parm))
{
co = which(names(object\$covariates) == parm[k])
partbasis = (sum(nb[0:(co-1)])+1):(sum(nb[0:co]))
if(center)
partbasis = partbasis+1

PB = c(PB,partbasis)
Koff = rbind(Koff,object\$coefficients[[co]])
}

if(center)
{
PB = c(1,PB)
Koff = rbind(object\$intercept,Koff)
}
B = object\$design[,PB,drop=FALSE]

fitti = B %*% Koff

for(i in 1:length(object\$asymmetries))
{
res[[i]] = matrix(NA,nrow=nrow(B),ncol=2)
colnames(res[[i]]) = c(paste(eval((1-level)/2),"%"),paste(eval((1+level)/2),"%"))
for(j in 1:nrow(B))
{
deviation = qnorm((1+level)/2) * sqrt(t(B[j,]) %*% object\$covmat[[i]][PB,PB] %*% B[j,])
res[[i]][j,] = c(fitti[j,i] - deviation, fitti[j,i] + deviation)
}
}
}

res
}
```

## Try the expectreg package in your browser

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

expectreg documentation built on March 18, 2022, 5:57 p.m.