search for: termmean

...ct, given the *name* of a term, which will always be a character string representing a factor ("A") or an interaction of two or more factors ("A:B", "A:B:C", ...). I'd prefer not to have to require() additional packages if possible. The functions I need are [1] termMeans(mod, term): Find the (possibly vector) response means for a model term, specified by a string, "A", "A:B", ... [2] dataIndices(mod, term): Find indices of the observations in the data corresponding to the combination of levels in the term Here is a small example to illustrat...

....0 4.798 8.100 0.2475 9.8575 11.970 Top:60-90 3.893 0.05775 1.4300 97.0 5.268 8.928 0.4250 9.2900 11.030 > Here is the current version of a function that doesn't work, because I can't supply the factor names to aggregate in the proper way. Can someone help me make it work? termMeans.mlm <- function( object, term ) { resp<- model.response(model.frame(object)) terms <- term.names(soils.mod) terms <- terms[terms != "(Intercept)"] factors <- strsplit(term, ":") # browser() means <- aggregate(resp, factors, mean) #...