[R-sig-ME] Data augmentation in TMB?
mollieebrooks at gmail.com
Tue Mar 13 17:49:36 CET 2018
This question would probably be better for the TMB users list tmb-users at googlegroups.com <mailto:tmb-users at googlegroups.com>. I just sent you an invitation to the group. If you are an absolute beginner with TMB, it may be more than the user list can help you with. Let me know and I’ll send you some resources.
You can fit the type of model you’re describing in TMB, but we will probably need more details. It’s fine that many observations are missing. From your question, I can’t tell if you have mathematically formulated the model you want to fit (e.g. distributions, process model). Or do you have some code already?
You could look at my state-space model of sheep growth for an example. It can show you the outline of the structure. Ignore the random effects section to begin with. You might just need a process model and 3 sets of observations.
Mollie E. Brooks, Ph.D.
National Institute of Aquatic Resources
Technical University of Denmark
> On 13Mar 2018, at 15:53, Marie RESCAN <marie.rescan at cefe.cnrs.fr> wrote:
> Dear R-sig-mixed-model users,
> I am currently fitting state space models using the TMB package and I
> would like to know if anyone have already implemented data augmentation
> in this package.
> My data consist of time series of 3 simultaneous types of observations
> of latent processes (dynamics of experimental populations of microalgae
> replicated in batch). I would like to fit a state space model on this
> data set, but many observations are missing:
> - For two observation methods, I cannot get population size estimation
> when the number of individuals is too low
> - For the last observation method, I got data only half of the time.
> I became aware that points in the latent process would not affect
> equally the when they do not have the same number of observations, which
> may be a problem to estimate process parameters.
> I would appreciate any suggestions to deal with this problem.
> Marie Rescan, PhD
> Post doctorante
> Centre d'écologie fonctionnelle et évolutive - CEFE
> UMR 5175
> Equipe Génétique & Ecologie évolutive - GEE
> 1919, route de Mende
> 34 293 Montpellier Cedex 5
> [[alternative HTML version deleted]]
> R-sig-mixed-models at r-project.org mailing list
[[alternative HTML version deleted]]
More information about the R-sig-mixed-models