This repo provides code related to the following two papers:
-
Foster, Hill and Lyons (2017) "Ecological Grouping of Survey Sites when Sampling Artefacts are Present". Journal of the Royal Statistical Society: Series C (Applied Statistics).
DOI: http://dx.doi.org/10.1111/rssc.12211 -
Lyons, Foster and Keith (2017, accepted) "Simultaneous vegetation classification and mapping at large spatial scales". Journal of Biogeography.
DOI: xx.xxxxx/xxx
The purpose of this repo is not necessarily to reproduce results form those papers (due to the amount of parallelised computing required), but to provide guidance for how analysis with the R package {RCPmod} can be parallelised on a (relatively generic) computing cluster. You would want (have) to do this when your analysis reaches 1000's of observations and 100's of species. The data is provided here as an .RData file for convenience. For some non-cluster work flows for fitting RCP models, see the repo https://github.com/mitchest/rcp-survey-artifacts
- The analysis is contained within the .R scripts
- The core fitting code is in "RCP2_katana.R", the other scripts are labelled more sensibly in terms of what they do =)
- The architechture specific paralellisation is found in the .pbs file (written for a Tourque cluster)
- I may or may not get rid of the hard-coded directories, sorry
- The best route for feed back on {RCPmod} is to lookup the function authors:
- https://cran.r-project.org/web/packages/RCPmod/index.html
- Otherwise, feel free to suggest edits here
- Mitchell Lyons ([email protected] / [email protected])