Skip to content
Prev 477 / 20628 Next

nlmer: gradient for random change point model

A completely (simpler and naive) approach could be to use an algorithm 
which works reasonably well in non-mixed GLM (see the package segmented).

The algorithm works via linear approximations of the kink function, 
therefore extension to mixed models is, at least in principle, 
straightforward. I discuss such an approach in a paper published in the 
Proceedings of the 19th IWSM, and I have (very) raw R code implementing 
the idea.

Let me know off-list if you are interested in the paper and/or code,

best,
vito


estimate the cluster-specific parameters (includ

Ken Beath ha scritto: