Species data segmentation for model run

[samaperrin]

Why have changes been made?

The species richness models aren't scaling well. Bob has suggested that we break the species up into smaller groups of 8-10 species, which can be run and then combined later. This change does so, creating groups of 10 species, which keep the 2 most common species (that are present in the prediction dataset) and then add 8 other species in. Additionally, there's a new column in the focalTaxa.csv file which allows us to dictate the predictionDataset that should be used.

What changes have been made?

• data/external/focalTaxa.csv - A new column has been added defining the prediction dataset for each group of species.
• functions/modelPreparation.R - This function (IF segmentation is set to TRUE) now divides up species into groups of 10, retaining the two most common species (that are also present in the prediction dataset) for comparison's sake. These become separate workflows, the results of which can be combined later.
• pipeline/models/speciesModelRuns.R - Segmentation is set to TRUE if we're measuring richness. The prediction dataset is also defined in the workflow.

[RRTogunov]

everything works up until specifying predictionIntercept in speciesModelRuns.R. I have never been able to get this to work... but segmentation seems fine, so I have approved and merged, but I haven't deleted the branch.

>   workflow$modelOptions(INLA = list(num.threads = 12, control.inla=list(int.strategy = 'eb', cmin = 0),safe = TRUE),
+                         Richness = list(predictionIntercept = predictionDatasetShort))
Error in workflow$modelOptions(INLA = list(num.threads = 12, control.inla = list(int.strategy = "eb",  : 
  predictionIntercept needs to be a name of one of the datasets in the model.