R package ctsem - hierarchical continuous time dynamic models


Hi all. As part of my phd I’ve been looking into dynamic models within developmental psychology, and working on some software to make it a bit easier for people to deal with. The software sets up a continuous (or discrete) time hierarchical state space model, using the Kalman filter (by default) to integrate over latent states. There is still some work needed, but it’s fairly functional as it stands, though definitely fails some of the posted guidelines - I’m not sure if they are all so applicable (avoiding compilation time when the models take much longer to fit anyway doesn’t seem so crucial) but will take a better look and change what I can sometime soon.

An intro to the package can be found at http://github.com/cdriveraus/ctsem/raw/master/vignettes/hierarchical.pdf
To get the very quick overview, looking at / running the sunspots example included in the intro vignette, starting bottom of page 11, might be simplest. If anyone is particularly interested I can also send through the soon to be submitted paper. I would of course be happy for any comments, on any aspect…

btw, the CRAN version is still only frequentist mixed effects.