Simulation from Multiple Fitted Models

[Ryan Martin]

Hey everyone,

I'm new to this group, but it seems like a great resource. I have a question as I begin to think about the modeling process. 

1. From reading through some of the old threads here, it appears the current implementation of the package only allows for prior segmentation of behavioral states/multiple home range centers and then independently fitting separate ctmm models to each subset of the data (please correct me if this assumption is wrong).

2. I am interested in simulating movement from the fitted models. However, it is unclear if that is easily facilitated through the current version ctmm::simulate() with multiple fitted models, or if I will need to custom code an implementation myself. Thanks for your time.

Best,

Ryan

[Christen Fleming]

Hi Ryan,

1. Yes.
2. We have done this in papers, such as https://movementecologyjournal.biomedcentral.com/articles/10.1186/s40462-019-0177-1, using simple tricks. For instance, if you just want to switch behavioral states at specific times, then you can simulate up to that time, and then run a second simulation conditioning off of the last point (though the velocity correlation would not persist through the transition). The code to do this more rigorously/seamlessly is mostly done and should be out next year.

Best,

Chris