1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
|
## Obtain fitted values from ssanova objects
fitted.ssanova <- function(object,...)
{
mf <- object$mf
if (!is.null(object$random)) mf$random <- I(object$random$z)
predict(object,mf)
}
## Obtain residuals from ssanova objects
residuals.ssanova <- function(object,...)
{
y <- model.response(object$mf,"numeric")
as.numeric(y-fitted.ssanova(object))
}
## Obtain fitted values in working scale from gssanova objects
fitted.gssanova <- function(object,...)
{
as.numeric(object$eta)
}
## Obtain residuals from gssanova objects
residuals.gssanova <- function(object,type="working",...)
{
y <- model.response(object$mf,"numeric")
wt <- model.weights(object$mf)
offset <- NULL
if ((object$family=="nbinomial")&(!is.null(object$nu))) y <- cbind(y,object$nu)
dat <- switch(object$family,
binomial=mkdata.binomial(y,object$eta,wt,offset),
nbinomial=mkdata.nbinomial(y,object$eta,wt,offset,object$nu),
poisson=mkdata.poisson(y,object$eta,wt,offset),
inverse.gaussian=mkdata.inverse.gaussian(y,object$eta,wt,offset),
Gamma=mkdata.Gamma(y,object$eta,wt,offset),
weibull=mkdata.weibull(y,object$eta,wt,offset,list(object$nu,FALSE)),
lognorm=mkdata.lognorm(y,object$eta,wt,offset,list(object$nu,FALSE)),
loglogis=mkdata.loglogis(y,object$eta,wt,offset,list(object$nu,FALSE)))
res <- as.numeric(dat$ywk - object$eta)
if (!is.na(charmatch(type,"deviance"))) {
dev.resid <- switch(object$family,
binomial=dev.resid.binomial(y,object$eta,wt),
nbinomial=dev.resid.nbinomial(y,object$eta,wt),
poisson=dev.resid.poisson(y,object$eta,wt),
inverse.gaussian=dev.resid.inverse.gaussian(y,object$eta,wt),
Gamma=dev.resid.Gamma(y,object$eta,wt),
weibull=dev.resid.weibull(y,object$eta,wt,object$nu),
lognorm=dev.resid.lognorm(y,object$eta,wt,object$nu),
loglogis=dev.resid.loglogis(y,object$eta,wt,object$nu))
res <- sqrt(dev.resid)*sign(res)
}
res
}
|
