Displaying 1 result from an estimated 1 matches for "dyba".
Did you mean:
dba
2011 Feb 15
0
Delta method using numerical derivatives
...uld like something akin to the following predictnl() function:
## get some data and fit a model
require(Epi)
data(lungDK)
fit <- glm(D/Y ~ as.factor(A5)+as.factor(A5):P5-1,
data=lungDK, subset=(P5>=1970), weight=Y,
family=poisson("identity")) # (Hakulinen and Dyba)
## calculate some summary statistics (e.g. cumulative incidence to 85
years)
require(plyr)
cuminc <- function(object,year=P5,newdata=NULL)
ddply(newdata,deparse(substitute(year)),function(data)
sum(5*predict(object,newdata=data)))[,-1,drop=TRUE]
## Now: get summary statistics and SEs...