Is it possible to include mixed effects into a random forest model in R? I know about the lmer (from lme4) and randomForest (from randomForest) functions but it would be nice if I could combine the two in some way, if that makes sense.
I'm afraid that if I don't include random effects in my random forest model, it will be incorrect.
The current main popular implementation of Random Forests (RF) (i.e. the randomForest package) is available only for univariate (continuous or discrete) responses. On the other hand, mixed models are inherently multivariate models, that is models that deal with vector-valued responses. Fortunately, extensions of RF for multivariate responses, in particular for handling longitudinal data, do exist.
LongitudiRF is one of the R packages that implement Random Forests for longitudinal data of which I am aware. A lot more information can be found at this recent review paper on longitudinal data with Random Forests.
Related posts:
How can I include random effects (or repeated measures) into a randomForest
How to deal with hierarchical / nested data in machine learning