Displaying 1 result from an estimated 1 matches for "weightexp".
Did you mean:
weighted
2007 May 23
0
Replicated LR goodness-of-fit tests, heterogeneity G, with loglm?
...xpi =
expcounts[tran==i]
G[i] =
2*sum((obsi*log(obsi/expi)))
}
dfs = rep(max(as.integer(levels(cov)))-1, max(as.integer(levels(cov))))
probs = pchisq(G, dfs, lower.tail=F) #and the lower tail probability of the G
#get a weighted
average for the pooled expectations
weightexp = list()
weights = vector()
for (i in
levels(tran)) {
expi = expprop[tran==i]
lengi =
tleng[as.integer(i)]
counti =
sum(obscounts[tran==i])
weightexp[[i]] =
expi*lengi*counti
weights[i] =
lengi*counti
}
sum = rep(0,
length(weightexp[[1...