Hi all, I am trying to use rpart.predict to test rpart models built with both numeric and character variables. The rpart trees appear to come out fine, but then when I try to use the rpart model to predict on a different dataset (same variables) I get error messages that claim the character variables are no good: "Warning messages: 1: In model.frame.default(Terms, newdata, na.action = na.action, xlev attr(object, : variable 'L3' is not a factor 2: In model.frame.default(Terms, newdata, na.action = na.action, xlev attr(object, : variable 'Range' is not a factor" When I build the original rpart models without these variables rpart.predict will work fine. Example of the R code I'm using: OVENrpart.pa <- rpart(OVENsites ~ L3 + BAavg + DBH + CACL + CanDecid + CanRich + CanDiv + SubcanCov + SubcanDecid + SubcanRich + SubcanDiv + Size + UpLow + Plantation + Range, data = sites1pabuild) OVEN.pa <- predict(OVENrpart.pa, newdata = sites1patest) "L3", "Plantation", and "Range" are all character variables, the rest are numeric. Thanks! Jay [[alternative HTML version deleted]]