This package aims at exploring the determinants of spatial data. It focuses on complex spatial patterns: influence of known barriers on the dispersion/spatial autocorrelation, multiple spatial scales, etc...

Simply source spatcontrol.R to use the spatial analysis functions:

source("spatcontrol.R")

It can ask for a bunch of R packages to be installed.

In addition, substantial computation time can be saved when using large datasets by using the attached C code:

R CMD SHLIB spatcontrol.c

More systematic use can be done by using the sec_launch.sh command:

chmod +x sec_launch.sh

Then you should use the example out of spatcontrol.c in a lower nivel than spatcontrol, calling it in the code (eg:mycode.R):

source("spatcontrol/spatcontrol.R,chdir=TRUE)

Then you can run a backed up simulation using: spatcontrol/sec_launch.sh mycode.R

Functions are in spatcontrol.R. Example of the use of the main functionalities are given the in the example_* files:

• how to generate spatially autocorrelated data in example_generation.R with gen.map
• how to compute structured autocorrelograms to examine the impact of known barriers on presence absence data in example_structuredMI.R.
• fit a gaussian field while accounting for known barriers, cofactors and observers quality in the fit in example_fit_GMRF.R The parameters for the priors are configured in parameters_extrapol.r

Other utility functions can be found in spatcontrol.R, organized into chapters:

• General purpose functions (data management, sparse matrices handling, plotting)
• Functions specific to the structured autocorrelograms
• Map generation
• Functions specific to the GMRF

This package is maintained by Corentin Barbu currently at University of Pennsylvania in MZ Levy lab.

The spatcontrol package is under development at the github repository: https://github.com/cbarbu/spatcontrol

Participation is welcome through forking and pull request in GitHub. The code will at some point be shared in part or fully as a CRAN package.

-Set more reasonable priors on io mean, io variance -Remove intercept from calculations carefully -Figure out where/what muPrior is doing (see parameters_extrapol.R) - and try to remove it (along with removing intercept)