SGAT (pronounced "tags backwards") provides facilities for estimating broadscale animal motions from archival or satellite tag data.
The package is easily built with RStudio
-
Install R
-
Install RStudio
-
Install Rtools or equivalent for your platform
-
Install devtools and roxygen2 packages and dependencies in R
-
Install geosphere, GeoLight, sp, raster, maptools packages and dependencies.
-
Clone the repository from GitHub (https://github.com/SWotherspoon/SGAT).
-
Create an Rstudio project in the folder containing this README file.
-
In the build tab, choose
More/Configure Build Tools...
and clickGenerate documentation with Roxygen
, selectConfigure
and choose to generateRd files
and theNAMESPACE file
, leaving the other options as they are. -
Choose
More/Roxygenize
from theBuild
tab -
Choose
Build & Reload
to make the package immediately available to R, or chooseMore/Build source package
More/Build binary package
from theBuild
tab to make source or binary packages.
-
Alternate Coordinates. The Metropolis sampler may be more efficient if locations are represented in geocentric ecliptic coordinates. But it is unclear whether the gains in efficiency due to higher acceptance rates would outweigh the costs of the coordinate transformation and the added code complexity. This is a low priority.
-
Parallelisation. At this point, the Metropolis samplers are only capable of utilizing a single core on a multicore machine. It would be relatively simple to introduce coarse grain parallelism by having the samplers draw multiple chains in parallel, using something like the multicore facility in the parallel package. Unfortunately, at the time of writing there does not seem to be a good parallization solution that works equally well on all platforms. This is a low priority.