Skip to content

predict GAMM

1 message · Rafael Wüest

#
Dear ME experts

I'm trying to use GAMMs in order to allow for a correlation structure in a GAM. (A simplified version of) My model looks 
as follows:

m1 <- gamm(y ~ s(p1) + s(p2), correlation = corSymm(form = ~ 1 | p3), data = data.me)

I have two related questions:

1) How do I get the fitted values of this model? I suspect I have to use predict(m1$lme) or fitted(m1$lme)...

2) Is it possible to predict with new data? predict(m1$gam) allows this via a newdata argument, but I suspect the 
correlation structure wouldn't be included then...

Many thanks for any hints,
Rafael