What I ultimately want to do is get y-hat scores after a multiple imputation, but Amelia does not provide fitted values. I have code that does it with a specific data set, but I'm trying to make a function that will do this regardless of the data set. Something like:
yhat<-function(a.out,num.obs,num.imp,model.qe){}
Where num.imp is the number of imputations used in amelia, or m. This is where I am stuck, though:
If a.out is an amelia object and names(a.out$imputations)[1] returns "imp1", why does a.out$imputations$names(a.out$imputations)[1] not return the same thing as a.out$imputations$"imp1"?
For a.out$imputations$names(a.out$imputations)[1], R says: Error: attempt to apply non-function.
How can I create a generic call for the individual imputations and the variables within?
Example taken from the Amelia docs
library(Amelia)
data(africa)
a.out <- amelia(x = africa, cs = "country", ts = "year", logs = "gdp_pc")
a.out$imputations is a list
str(a.out$imputations)
Elements in a list can be called either by name or numeric index
head(a.out$imputations$imp1)
head(a.out$imputations[["imp1"]])
head(a.out$imputations[[1]])
Each imputation is a data frame. A data frame is just a particular kind of list and elements can be called in that manner.
head(a.out$imputations[[1]]$country)
head(a.out$imputations[[1]][["country"]])
head(a.out$imputations[[1]][[2]])
head(a.out$imputations[[1]][2])
The last of those methods differ from the rest in that it returns a one-column data frame, while the others return a vector.
To extract an individual value entirely by numeric indices you could do f.ex
a.out$imputations[[1]][3, 2]
or
a.out$imputations[[1]][[2]][3]