Nothing
R/sdmWorkflow.R
In intSDM: Reproducible Integrated Species Distribution Models Across Norway using 'INLA'
Defines functions
sdmWorkflow
Documented in
sdmWorkflow
#' @title \code{sdmWorkflow}: Function to compile the reproducible workflow.
#' @description This function is used to compile the reproducible workflow from the \code{R6} object created with \code{startFunction}. Depending on what was specified before, this function will estimate the integrated species distribution model, perform cross-validation, create predictions from the model and plot these predictions.
#' @param Workflow The \code{R6} object created from \code{startWorkflow}. This object should contain all the data and model information required to estimate and specify the model.
#' @param predictionDim The pixel dimensions for the prediction maps. Defaults to \code{c(150, 150)}.
#' @param predictionData Optional argument for the user to specify their own data to predict on. Must be a \code{sf} or \code{SpatialPixelsDataFrame} object. Defaults to \code{NULL}.
#' @param initialValues Find initial values using a GLM before the model is estimated. Defaults to \code{FALSE}.
#' @param inlaOptions Options to specify in \link[INLA]{inla} from the \code{inla} function. See \code{?inla} for more details.
#' @param ipointsOptions Options to specify in \link[inlabru]{fm_int}'s \code{int.args} argument. See \code{?fmesher::fm_int} for more details.
#'
#' @import PointedSDMs
#'
#' @return The return of the function depends on the argument \code{Save} from the \code{startWorkflow} function. If this argument is \code{FALSE} then the objects will be saved to the specidfied directory. If this argument is \code{TRUE} then a list of different outcomes from the workflow will be returned.
#' @export
#' @examples
#' \dontrun{
#' if (requireNamespace('INLA')) {
#'
#' workflow <- startWorkflow(Species = 'Fraxinus excelsior',
#' Projection = '+proj=longlat +ellps=WGS84',
#' Save = FALSE,
#' saveOptions = list(projectName = 'example'))
#' workflow$addArea(countryName = 'Sweden')
#'
#' workflow$addGBIF(datasetName = 'exampleGBIF',
#' datasetType = 'PA',
#' limit = 10000,
#' coordinateUncertaintyInMeters = '0,50')
#' workflow$addMesh(cutoff = 20000,
#' max.edge=c(60000, 80000),
#' offset= 100000)
#' workflow$workflowOutput('Model')
#'
#' Model <- sdmWorkflow(workflow)
#'
#' }
#' }
sdmWorkflow <- function(Workflow = NULL,
predictionDim = c(150, 150),
predictionData = NULL,
initialValues = FALSE,
inlaOptions = list(),
ipointsOptions = NULL) {
modDirectory <- Workflow$.__enclos_env__$private$Directory
saveObjects <- Workflow$.__enclos_env__$private$Save
Quiet <- Workflow$.__enclos_env__$private$Quiet
if (is.null(Workflow$.__enclos_env__$private$Mesh)) stop('An fm_mesh_2d object is required before any analysis is completed. Please add using the `.$addMesh` function.')
if (is.null(Workflow$.__enclos_env__$private$Output)) stop('A model output needs to be specified before any analysis is completed. Please add using the `.$workflowOutput` function.')
Oputs <- Workflow$.__enclos_env__$private$Output
outputList <-list()
if ('Bias' %in% Oputs &&
all(is.null(Workflow$.__enclos_env__$private$biasNames),
is.null(Workflow$.__enclos_env__$private$biasFormula))) stop('Bias specified as output but no bias fields were added. Please do this with .$biasFields')
if (is.null(Workflow$.__enclos_env__$private$CVMethod) && 'Cross-Validation' %in% Workflow$.__enclos_env__$private$Output) stop('Cross-validation specified as model output but no method provided. Please specify cross-validation method using the `.$crossValidation` function.')
##Need to do it on a species by species basis:
if (length(Workflow$.__enclos_env__$private$optionsISDM) > 0) {
.__pointsSpatial.__ <- Workflow$.__enclos_env__$private$optionsISDM$pointsSpatial
if (!is.null(Workflow$.__enclos_env__$private$optionsISDM[['pointsIntercept']])) .__pointsIntercept.__ <- Workflow$.__enclos_env__$private$optionsISDM$pointsIntercept
else .__pointsIntercept.__ <- TRUE
if (!is.null(Workflow$.__enclos_env__$private$optionsISDM[['pointCovariates']])) .__pointCovariates.__ <- Workflow$.__enclos_env__$private$optionsISDM$pointsIntercept
else .__pointCovariates.__ <- NULL
if (!is.null(Workflow$.__enclos_env__$private$optionsISDM[['Offset']])) .__Offset.__ <- Workflow$.__enclos_env__$private$optionsISDM$Offset
else .__Offset.__ <- NULL
} else {
if (!Workflow$.__enclos_env__$private$richnessEstimate) .__pointsSpatial.__ <- 'copy'
else .__pointsSpatial.__ <- NULL
.__pointsIntercept.__ <- TRUE
.__Offset.__ <- NULL
.__pointCovariates.__ <- NULL
}
if (!is.null(Workflow$.__enclos_env__$private$copyModel)) .__copyModel.__ <- Workflow$.__enclos_env__$private$copyModel
else .__copyModel.__ <- deparse1('list(beta = list(fixed = FALSE))')
.__mesh.__ <- Workflow$.__enclos_env__$private$Mesh
.__proj.__ <- Workflow$.__enclos_env__$private$Projection
.__coordinates.__ <- Workflow$.__enclos_env__$private$Coordinates
.__responsePA.__ <- Workflow$.__enclos_env__$private$responsePA
.__responseCounts.__ <- Workflow$.__enclos_env__$private$responseCounts
.__trialsName.__ <- Workflow$.__enclos_env__$private$trialsName
if (length(Workflow$.__enclos_env__$private$Covariates) > 0) {
covRes <- sapply(Workflow$.__enclos_env__$private$Covariates, function(x) res(x)[1])
if (length(unique(covRes)) > 1) {
lowRes <- names(which.max(covRes))
spatCovs <- terra::rast(lapply(Workflow$.__enclos_env__$private$Covariates, FUN = function(x) resample(x, Workflow$.__enclos_env__$private$Covariates[[lowRes]])))
} else spatCovs <- terra::rast(Workflow$.__enclos_env__$private$Covariates)
}
else spatCovs <- NULL
spatCovs <- do.call(c, unlist(list(spatCovs, Workflow$.__enclos_env__$private$biasCovariates), recursive = FALSE))
IPS <- fm_int(domain = .__mesh.__, samplers = Workflow$.__enclos_env__$private$Area,
int.args = ipointsOptions)
st_geometry(IPS) <- 'geometry'
IPS <- IPS[sapply(st_intersects(IPS, Workflow$.__enclos_env__$private$Area), function(z) if (length(z)==0) FALSE else TRUE),]
if (!Workflow$.__enclos_env__$private$richnessEstimate) {
for (species in unique(c(names(Workflow$.__enclos_env__$private$dataGBIF),
names(Workflow$.__enclos_env__$private$dataStructured)))) {
speciesNameInd <- sub(' ', '_', species)
if (saveObjects) dir.create(path = paste0(modDirectory, '/', speciesNameInd))
speciesDataset <- append(Workflow$.__enclos_env__$private$dataGBIF[[species]],
Workflow$.__enclos_env__$private$dataStructured[[species]])
speciesDataset <- speciesDataset[sapply(speciesDataset, nrow) > 0]
if (length(speciesDataset) == 0) {
warning(paste('No data added to the model for species', paste0(species,'.'), 'Skipping run.'))
}
else {
if (!Quiet) {
message(paste('\nStarting model for', species, '\n\n'))
message('\nInitializing model', '\n\n')
}
if (length(speciesDataset) == 1) {
initializeModel <- specifyISDM$new(data = speciesDataset,
projection = .__proj.__,
Inlamesh = .__mesh.__,
initialnames = names(speciesDataset),
responsecounts = .__responseCounts.__,
responsepa = .__responsePA.__,
pointcovariates = .__pointCovariates.__,
trialspa = .__trialsName.__,
spatial = 'individual',
intercepts = .__pointsIntercept.__,
spatialcovariates = spatCovs,
boundary = Workflow$.__enclos_env__$private$Area,
ips = IPS,
temporal = NULL,
temporalmodel = NULL,
offset = .__Offset.__,
copymodel = .__copyModel.__,
formulas = list(covariateFormula = Workflow$.__enclos_env__$private$covariateFormula,
biasFormula = Workflow$.__enclos_env__$private$biasFormula))
}
else {
initializeModel <- PointedSDMs::startISDM(speciesDataset, Mesh = .__mesh.__, Projection = .__proj.__,
responsePA = .__responsePA.__, responseCounts = .__responseCounts.__,
trialsPA = .__trialsName.__, pointsSpatial = .__pointsSpatial.__,
pointsIntercept = .__pointsIntercept.__ , IPS = IPS,
Offset = .__Offset.__,
pointCovariates = .__pointCovariates.__,
Boundary = Workflow$.__enclos_env__$private$Area,
spatialCovariates = spatCovs,
Formulas = list(covariateFormula = Workflow$.__enclos_env__$private$covariateFormula,
biasFormula = Workflow$.__enclos_env__$private$biasFormula))
}
if (!is.null(Workflow$.__enclos_env__$private$priorsFixed)) {
for (var in names(Workflow$.__enclos_env__$private$priorsFixed)) {
initializeModel$priorsFixed(Effect = var, mean.linear = Workflow$.__enclos_env__$private$priorsFixed[[var]][1], prec.linear = Workflow$.__enclos_env__$private$priorsFixed[[var]][2])
}
}
##Here priors for the random effects
if (!is.null(Workflow$.__enclos_env__$private$copyModel)) initializeModel$specifyRandom(copyModel = .__copyModel.__)
#Workflow$specifyRandom(copyModel = x)
if (!is.null(.__pointsSpatial.__)) {
if (!is.null(Workflow$.__enclos_env__$private$sharedField)) {
if (.__pointsSpatial.__ %in% c('shared', 'correlate')) initializeModel$spatialFields$sharedField$sharedField <- Workflow$.__enclos_env__$private$sharedField
else {
for (data in names(initializeModel$spatialFields$datasetFields)) {
initializeModel$spatialFields$datasetFields[[data]] <- Workflow$.__enclos_env__$private$sharedField
}
}
}
}
if (!is.null(Workflow$.__enclos_env__$private$biasNames)) {
if (!any(names(speciesDataset) %in% Workflow$.__enclos_env__$private$biasNames)) {
warning('Bias fields specified for datasets not in the model. Turning bias off.')
biasIn <- FALSE
}
else {
biasIn <- TRUE
biasSubset <- Workflow$.__enclos_env__$private$biasNames[Workflow$.__enclos_env__$private$biasNames %in% names(speciesDataset)]
initializeModel$addBias(datasetNames = biasSubset, copyModel = Workflow$.__enclos_env__$private$biasFieldsCopy, shareModel = Workflow$.__enclos_env__$private$biasFieldsShare)
if (!is.null(Workflow$.__enclos_env__$private$biasFieldsSpecify)) {
if (any(names(speciesDataset) %in% names(Workflow$.__enclos_env__$private$biasFieldsSpecify))) {
specifySubset <- names(Workflow$.__enclos_env__$private$biasFieldsSpecify)[names(Workflow$.__enclos_env__$private$biasFieldsSpecify) %in% names(speciesDataset)]
for (biasName in specifySubset) {
initializeModel$spatialFields$biasFields[[biasName]] <- Workflow$.__enclos_env__$private$biasFieldsSpecify[[biasName]]
}
}
}
}
} else biasIn <- FALSE
if (!is.null(Workflow$.__enclos_env__$private$biasFormula)) biasIn <- TRUE
else
if (!biasIn) biasIn <- FALSE
else biasIn <- TRUE
if ('Cross-validation' %in% Oputs && 'spatialBlock' %in%Workflow$.__enclos_env__$private$CVMethod) {
initializeModel$spatialBlock(k = Workflow$.__enclos_env__$private$blockOptions$k,
rows_cols = Workflow$.__enclos_env__$private$blockOptions$rows_cols,
seed = Workflow$.__enclos_env__$private$blockOptions$seed)
}
if (initialValues) Workflow$.__enclos_env__$private$optionsINLA[['bru_initial']] <- initValues(data = initializeModel, formulaComponents = initializeModel$.__enclos_env__$private$spatcovsNames)
message('\nEstimating ISDM:\n\n')
PSDMsMOdel <- try(PointedSDMs::fitISDM(initializeModel,
options = inlaOptions))
if (inherits(PSDMsMOdel, 'try-error')) warning(paste0('Model estimation failed for ', species,'. Will skip the rest of the outputs.'))
if ('Model' %in% Oputs) {
if (saveObjects) {
if (!Quiet) message('\nSaving Model object:', '\n\n')
saveRDS(object = PSDMsMOdel, file = paste0(modDirectory,'/', speciesNameInd, '/intModel.rds'))
} else outputList[[speciesNameInd]][['Model']] <- PSDMsMOdel
}
if ('Summary' %in% Oputs) {
if ('Fixed__Effects__Comps' %in% names(PSDMsMOdel$summary.random)) {
fixedRandom <-PSDMsMOdel$summary.random$Fixed__Effects__Comps
row.names(fixedRandom) <- fixedRandom$ID
fixedRandom$ID <- NULL
} else fixedRandom <- data.frame()
if ('Bias__Effects__Comps' %in% names(PSDMsMOdel$summary.random)) {
biasFixed <-PSDMsMOdel$summary.random$Bias__Effects__Comps
row.names(biasFixed) <- biasFixed$ID
biasFixed$ID <- NULL
} else biasFixed <- data.frame()
summariesIndex <- list(Fixed = rbind(PSDMsMOdel$summary.fixed, fixedRandom, biasFixed),
Hyper = PSDMsMOdel$summary.hyperpar)
if (saveObjects) {
if (!Quiet) message('\nSaving Model summaries:', '\n\n')
saveRDS(object = summariesIndex, file = paste0(modDirectory,'/', speciesNameInd, '/modelSummary.rds'))
} else outputList[[speciesNameInd]][['Summary']] <- summariesIndex
}
if ('Cross-validation' %in% Oputs && !inherits(PSDMsMOdel, 'try-error')) {
if ('spatialBlock' %in% Workflow$.__enclos_env__$private$CVMethod) {
if (!Quiet) message('\nEstimating spatial block cross-validation:\n\n')
if (Workflow$.__enclos_env__$private$blockCVType == 'DIC') spatialBlockCV <- PointedSDMs::blockedCV(initializeModel, options = inlaOptions)
else {
blockCVPredName <- names(PSDMsMOdel$dataType)[PSDMsMOdel$dataType != 'Present only'][1]
spatialBlockCV <- PointedSDMs::blockedCV(initializeModel, options = inlaOptions, method = 'Predict', predictName = blockCVPredName)
}
if (saveObjects) {
if (!Quiet) message('\nSaving spatial blocked cross-validation object:', '\n\n')
saveRDS(object = spatialBlockCV, file = paste0(modDirectory,'/', speciesNameInd, '/spatialBlock.rds'))
} else outputList[[speciesNameInd]][['spatialBlock']] <- spatialBlockCV
rm(spatialBlockCV)
}
if ('Loo' %in% Workflow$.__enclos_env__$private$CVMethod && !inherits(PSDMsMOdel, 'try-error')) {
if (!Quiet) message('\nEstimating leave-one-out cross-validation:\n\n')
LooCV <- PointedSDMs::datasetOut(model = PSDMsMOdel)
if (saveObjects) {
if (!Quiet) message('\nSaving leave-one-out cross-validation object:', '\n\n')
saveRDS(object = LooCV, file = paste0(modDirectory,'/', speciesNameInd, '/LooCV.rds'))
} else outputList[[speciesNameInd]][['LooCV']] <- LooCV
rm(LooCV)
}
}
if (any(c('Predictions', 'Maps') %in% Oputs) && !inherits(PSDMsMOdel, 'try-error')) {
if (!Quiet) message('\nPredicting model:\n\n')
if (is.null(predictionData)) {
.__mask.__ <- as(Workflow$.__enclos_env__$private$Area, 'Spatial')
predictionData <- fmesher::fm_pixels(mesh = .__mesh.__,
mask = .__mask.__,
dims = predictionDim)
}
Predictions <- predict(PSDMsMOdel, data = predictionData, predictor = TRUE)
if (saveObjects) {
if (!Quiet) message('\nSaving predictions object:', '\n\n')
saveRDS(object = Predictions, file = paste0(modDirectory,'/', speciesNameInd, '/Predictions.rds'))
} else outputList[[speciesNameInd]][['Predictions']] <- Predictions
}
if ('Maps' %in% Oputs && !inherits(PSDMsMOdel, 'try-error')) {
if (!Quiet) message("\nPlotting Maps:\n\n")
if (saveObjects) {
if (!Quiet) message('\nSaving plots object:', '\n\n')
plot(Predictions)
ggsave(filename = paste0(modDirectory,'/', speciesNameInd, '/Map.png'))
}
else outputList[[speciesNameInd]][['Maps']] <- plot(Predictions, plot = FALSE)
rm(Predictions)
}
if ('Bias' %in% Oputs && !inherits(PSDMsMOdel, 'try-error')) {
if (biasIn) {
if (!Quiet) message('\nProducing bias predictions:\n\n')
if (is.null(predictionData)) {
.__mask.__ <- as(Workflow$.__enclos_env__$private$Area, 'Spatial')
predictionData <- fmesher::fm_pixels(mesh = .__mesh.__,
mask = .__mask.__,
dims = predictionDim)
}
biasPreds <- predict(PSDMsMOdel,
data = predictionData,
bias = TRUE)
if (saveObjects) {
if (!Quiet) message('\nSaving predictions object:', '\n\n')
saveRDS(object = biasPreds, file = paste0(modDirectory,'/', speciesNameInd, '/biasPreds.rds'))
} else outputList[[speciesNameInd]][['Bias']] <- biasPreds
}
}
}
}
}
else {
if (is.null(Workflow$.__enclos_env__$private$optionsRichness[['predictionIntercept']])) stop('predictionIntercept needs to be provided. This can be done using .$modelOptions(Richness = list(predictionIntercept = "DATASETNAME")).')
.__predIntercept.__ <- paste0(Workflow$.__enclos_env__$private$optionsRichness[['predictionIntercept']],'_intercept')
.__spatModel.__ <- Workflow$.__enclos_env__$private$optionsRichness[['speciesSpatial']]
##Fix this
#Need to get the datasets back together
spNames <- c(names(Workflow$.__enclos_env__$private$dataGBIF),
names(Workflow$.__enclos_env__$private$dataStructured))
spData <- append(unlist(Workflow$.__enclos_env__$private$dataGBIF, recursive = FALSE),
unlist(Workflow$.__enclos_env__$private$dataStructured, recursive = FALSE))
names(spData) <- spNames
wMain <- which(spNames == Workflow$.__enclos_env__$private$optionsRichness[['predictionIntercept']])
spData <- spData[c(spNames[wMain], spNames[-wMain])]
message('Setting up richness model:', '\n\n')
richSetup <- PointedSDMs::startSpecies(spData, Mesh = .__mesh.__, Projection = .__proj.__,
responsePA = .__responsePA.__, responseCounts = .__responseCounts.__,
trialsPA = .__trialsName.__, Boundary = Workflow$.__enclos_env__$private$Area,
pointsIntercept = .__pointsIntercept.__,
IPS = IPS,
pointCovariates = .__pointCovariates.__,
Offset = .__Offset.__,
speciesIntercept = Workflow$.__enclos_env__$private$speciesIntercept,
speciesName = Workflow$.__enclos_env__$private$speciesName,
speciesSpatial = .__spatModel.__,
pointsSpatial = .__pointsSpatial.__, # Make this an argument
spatialCovariates = spatCovs,
Formulas = list(covariateFormula = Workflow$.__enclos_env__$private$covariateFormula,
biasFormula = Workflow$.__enclos_env__$private$biasFormula))
##Redo this
if (!is.null(Workflow$.__enclos_env__$private$priorsFixed)) {
for (var in names(Workflow$.__enclos_env__$private$priorsFixed)) {
richSetup$priorsFixed(Effect = var, mean.linear = Workflow$.__enclos_env__$private$priorsFixed[[var]][1], prec.linear = Workflow$.__enclos_env__$private$priorsFixed[[var]][2])
}
}
if (!is.null(Workflow$.__enclos_env__$private$sharedField)) {
richSetup$spatialFields$speciesFields$speciesField <- Workflow$.__enclos_env__$private$sharedField
if (!is.null(.__pointsSpatial.__)) {
if (.__pointsSpatial.__ == 'copy') {
for (dtt in names(richSetup$spatialFields$datasetFields)) {
richSetup$spatialFields$datasetFields[[dtt]] <- Workflow$.__enclos_env__$private$sharedField
}
}
}
}
#Add copy here use specifySpatial?
if (!is.null(Workflow$.__enclos_env__$private$biasNames)) {
richSetup$addBias(datasetNames = Workflow$.__enclos_env__$private$biasNames, copyModel = Workflow$.__enclos_env__$private$biasFieldsCopy, shareModel = Workflow$.__enclos_env__$private$biasFieldsShare)
if (!is.null(Workflow$.__enclos_env__$private$biasFieldsSpecify)) {
for (biasName in names(Workflow$.__enclos_env__$private$biasFieldsSpecify)) {
richSetup$spatialFields$biasFields[[biasName]] <- Workflow$.__enclos_env__$private$biasFieldsSpecify[[biasName]]
}
}
}
#Redo this with specifyRandom
if (!is.null(Workflow$.__enclos_env__$private$priorIntercept)) richSetup$specifyRandom(speciesIntercepts = Workflow$.__enclos_env__$private$priorIntercept)
if (!is.null(Workflow$.__enclos_env__$private$optionsRichness$speciesSpatial)) {
if (!is.null(Workflow$.__enclos_env__$private$priorGroup) && Workflow$.__enclos_env__$private$optionsRichness$speciesSpatial == 'replicate') richSetup$specifyRandom(speciesGroup = Workflow$.__enclos_env__$private$priorGroup)
}
if (initialValues) Workflow$.__enclos_env__$private$optionsINLA[['bru_initial']] <- initValues(data = richSetup, formulaComponents = richSetup$.__enclos_env__$private$spatcovsNames)
message('Estimating richness model:', '\n\n')
richModel <- try(PointedSDMs::fitISDM(data = richSetup,
options = inlaOptions))
if (inherits(richModel, 'try-error')) stop('Richness model failed to estimate. Will skip the rest of the outputs.')
else
if ('Model' %in% Oputs) {
if (saveObjects) {
if (!Quiet) message('\nSaving richness model:', '\n\n')
saveRDS(object = richModel, file = paste0(modDirectory, '/richnessModel.rds')) #Add project name here
} else outputList[['RichnessModel']] <- richModel
}
if (is.null(predictionData)) {
.__mask.__ <- as(Workflow$.__enclos_env__$private$Area, 'sf')
predictionData <- fmesher::fm_pixels(mesh = .__mesh.__,
mask = .__mask.__,
dims = predictionDim)
}
if (any(c('Predictions', 'Maps') %in% Oputs) && !inherits(richModel, 'try-error')) {
richOutput <- obtainRichness(modelObject = richModel, predictionData = predictionData,
predictionIntercept = Workflow$.__enclos_env__$private$optionsRichness[['predictionIntercept']],
sampleSize = Workflow$.__enclos_env__$private$samplingSize)
if (saveObjects) {
if (!Quiet) message('\nSaving richness predictions:', '\n\n')
saveRDS(object = richOutput, file = paste0(modDirectory, '/richnessPredictions.rds')) #Add project name here
} else outputList[['Richness']] <- richOutput#richPredicts
}
#}
if ('Summary' %in% Oputs) {
removeList <- grepl('spatial', names(richModel$summary.random)) | names(richModel$summary.random) == 'speciesShared'
if (paste0(richModel$species$speciesVar,'_intercepts') %in% names(richModel$summary.random)) richModel$summary.random[[paste0(richModel$species$speciesVar,'_intercepts')]]$ID <- paste0(row.names(richModel$summary.random[[paste0(richModel$species$speciesVar,'_intercepts')]]), '_intercept')
richnessSummary <- list(Fixed = richModel$summary.fixed,
Random = do.call(rbind, richModel$summary.random[!removeList]),
Hyperparameters = richModel$summary.hyperpar)
row.names(richnessSummary$Random) <- NULL
if (saveObjects) {
if (!Quiet) message('\nSaving richness summaries:', '\n\n')
saveRDS(object = richnessSummary, file = paste0(modDirectory, '/richnessSummaries.rds')) #Add project name here
} else outputList[['Summary']] <- richnessSummary
}
if ('Bias' %in% Oputs) {
if (!Quiet) message('\nProducing bias predictions:\n\n')
if (is.null(predictionData)) {
.__mask.__ <- as(Workflow$.__enclos_env__$private$Area, 'Spatial')
predictionData <- fmesher::fm_pixels(mesh = .__mesh.__,
mask = .__mask.__,
dims = predictionDim)
}
biasPreds <- predict(richModel,
data = predictionData,
bias = TRUE)
if (saveObjects) {
if (!Quiet) message('\nSaving predictions object:', '\n\n')
saveRDS(object = biasPreds, file = paste0(modDirectory,'/', '/biasRichnessPreds.rds')) #Add project name here
} else outputList[['BiasRichness']] <- biasPreds
}
}
if (!saveObjects) {
if (!exists('outputList')) stop('Workflow did not run for any species.')
return(outputList)
}
}
Try the intSDM package in your browser
Any scripts or data that you put into this service are public.
intSDM documentation built on Sept. 9, 2025, 5:59 p.m.
R Package Documentation
Browse R Packages
We want your feedback!
Note that we can't provide technical support on individual packages. You should contact the package authors for that.
Defines functions sdmWorkflow
Documented in sdmWorkflow
#' @title \code{sdmWorkflow}: Function to compile the reproducible workflow.
#' @description This function is used to compile the reproducible workflow from the \code{R6} object created with \code{startFunction}. Depending on what was specified before, this function will estimate the integrated species distribution model, perform cross-validation, create predictions from the model and plot these predictions.
#' @param Workflow The \code{R6} object created from \code{startWorkflow}. This object should contain all the data and model information required to estimate and specify the model.
#' @param predictionDim The pixel dimensions for the prediction maps. Defaults to \code{c(150, 150)}.
#' @param predictionData Optional argument for the user to specify their own data to predict on. Must be a \code{sf} or \code{SpatialPixelsDataFrame} object. Defaults to \code{NULL}.
#' @param initialValues Find initial values using a GLM before the model is estimated. Defaults to \code{FALSE}.
#' @param inlaOptions Options to specify in \link[INLA]{inla} from the \code{inla} function. See \code{?inla} for more details.
#' @param ipointsOptions Options to specify in \link[inlabru]{fm_int}'s \code{int.args} argument. See \code{?fmesher::fm_int} for more details.
#'
#' @import PointedSDMs
#'
#' @return The return of the function depends on the argument \code{Save} from the \code{startWorkflow} function. If this argument is \code{FALSE} then the objects will be saved to the specidfied directory. If this argument is \code{TRUE} then a list of different outcomes from the workflow will be returned.
#' @export
#' @examples
#' \dontrun{
#' if (requireNamespace('INLA')) {
#'
#' workflow <- startWorkflow(Species = 'Fraxinus excelsior',
#' Projection = '+proj=longlat +ellps=WGS84',
#' Save = FALSE,
#' saveOptions = list(projectName = 'example'))
#' workflow$addArea(countryName = 'Sweden')
#'
#' workflow$addGBIF(datasetName = 'exampleGBIF',
#' datasetType = 'PA',
#' limit = 10000,
#' coordinateUncertaintyInMeters = '0,50')
#' workflow$addMesh(cutoff = 20000,
#' max.edge=c(60000, 80000),
#' offset= 100000)
#' workflow$workflowOutput('Model')
#'
#' Model <- sdmWorkflow(workflow)
#'
#' }
#' }
sdmWorkflow <- function(Workflow = NULL,
predictionDim = c(150, 150),
predictionData = NULL,
initialValues = FALSE,
inlaOptions = list(),
ipointsOptions = NULL) {
modDirectory <- Workflow$.__enclos_env__$private$Directory
saveObjects <- Workflow$.__enclos_env__$private$Save
Quiet <- Workflow$.__enclos_env__$private$Quiet
if (is.null(Workflow$.__enclos_env__$private$Mesh)) stop('An fm_mesh_2d object is required before any analysis is completed. Please add using the `.$addMesh` function.')
if (is.null(Workflow$.__enclos_env__$private$Output)) stop('A model output needs to be specified before any analysis is completed. Please add using the `.$workflowOutput` function.')
Oputs <- Workflow$.__enclos_env__$private$Output
outputList <-list()
if ('Bias' %in% Oputs &&
all(is.null(Workflow$.__enclos_env__$private$biasNames),
is.null(Workflow$.__enclos_env__$private$biasFormula))) stop('Bias specified as output but no bias fields were added. Please do this with .$biasFields')
if (is.null(Workflow$.__enclos_env__$private$CVMethod) && 'Cross-Validation' %in% Workflow$.__enclos_env__$private$Output) stop('Cross-validation specified as model output but no method provided. Please specify cross-validation method using the `.$crossValidation` function.')
##Need to do it on a species by species basis:
if (length(Workflow$.__enclos_env__$private$optionsISDM) > 0) {
.__pointsSpatial.__ <- Workflow$.__enclos_env__$private$optionsISDM$pointsSpatial
if (!is.null(Workflow$.__enclos_env__$private$optionsISDM[['pointsIntercept']])) .__pointsIntercept.__ <- Workflow$.__enclos_env__$private$optionsISDM$pointsIntercept
else .__pointsIntercept.__ <- TRUE
if (!is.null(Workflow$.__enclos_env__$private$optionsISDM[['pointCovariates']])) .__pointCovariates.__ <- Workflow$.__enclos_env__$private$optionsISDM$pointsIntercept
else .__pointCovariates.__ <- NULL
if (!is.null(Workflow$.__enclos_env__$private$optionsISDM[['Offset']])) .__Offset.__ <- Workflow$.__enclos_env__$private$optionsISDM$Offset
else .__Offset.__ <- NULL
} else {
if (!Workflow$.__enclos_env__$private$richnessEstimate) .__pointsSpatial.__ <- 'copy'
else .__pointsSpatial.__ <- NULL
.__pointsIntercept.__ <- TRUE
.__Offset.__ <- NULL
.__pointCovariates.__ <- NULL
}
if (!is.null(Workflow$.__enclos_env__$private$copyModel)) .__copyModel.__ <- Workflow$.__enclos_env__$private$copyModel
else .__copyModel.__ <- deparse1('list(beta = list(fixed = FALSE))')
.__mesh.__ <- Workflow$.__enclos_env__$private$Mesh
.__proj.__ <- Workflow$.__enclos_env__$private$Projection
.__coordinates.__ <- Workflow$.__enclos_env__$private$Coordinates
.__responsePA.__ <- Workflow$.__enclos_env__$private$responsePA
.__responseCounts.__ <- Workflow$.__enclos_env__$private$responseCounts
.__trialsName.__ <- Workflow$.__enclos_env__$private$trialsName
if (length(Workflow$.__enclos_env__$private$Covariates) > 0) {
covRes <- sapply(Workflow$.__enclos_env__$private$Covariates, function(x) res(x)[1])
if (length(unique(covRes)) > 1) {
lowRes <- names(which.max(covRes))
spatCovs <- terra::rast(lapply(Workflow$.__enclos_env__$private$Covariates, FUN = function(x) resample(x, Workflow$.__enclos_env__$private$Covariates[[lowRes]])))
} else spatCovs <- terra::rast(Workflow$.__enclos_env__$private$Covariates)
}
else spatCovs <- NULL
spatCovs <- do.call(c, unlist(list(spatCovs, Workflow$.__enclos_env__$private$biasCovariates), recursive = FALSE))
IPS <- fm_int(domain = .__mesh.__, samplers = Workflow$.__enclos_env__$private$Area,
int.args = ipointsOptions)
st_geometry(IPS) <- 'geometry'
IPS <- IPS[sapply(st_intersects(IPS, Workflow$.__enclos_env__$private$Area), function(z) if (length(z)==0) FALSE else TRUE),]
if (!Workflow$.__enclos_env__$private$richnessEstimate) {
for (species in unique(c(names(Workflow$.__enclos_env__$private$dataGBIF),
names(Workflow$.__enclos_env__$private$dataStructured)))) {
speciesNameInd <- sub(' ', '_', species)
if (saveObjects) dir.create(path = paste0(modDirectory, '/', speciesNameInd))
speciesDataset <- append(Workflow$.__enclos_env__$private$dataGBIF[[species]],
Workflow$.__enclos_env__$private$dataStructured[[species]])
speciesDataset <- speciesDataset[sapply(speciesDataset, nrow) > 0]
if (length(speciesDataset) == 0) {
warning(paste('No data added to the model for species', paste0(species,'.'), 'Skipping run.'))
}
else {
if (!Quiet) {
message(paste('\nStarting model for', species, '\n\n'))
message('\nInitializing model', '\n\n')
}
if (length(speciesDataset) == 1) {
initializeModel <- specifyISDM$new(data = speciesDataset,
projection = .__proj.__,
Inlamesh = .__mesh.__,
initialnames = names(speciesDataset),
responsecounts = .__responseCounts.__,
responsepa = .__responsePA.__,
pointcovariates = .__pointCovariates.__,
trialspa = .__trialsName.__,
spatial = 'individual',
intercepts = .__pointsIntercept.__,
spatialcovariates = spatCovs,
boundary = Workflow$.__enclos_env__$private$Area,
ips = IPS,
temporal = NULL,
temporalmodel = NULL,
offset = .__Offset.__,
copymodel = .__copyModel.__,
formulas = list(covariateFormula = Workflow$.__enclos_env__$private$covariateFormula,
biasFormula = Workflow$.__enclos_env__$private$biasFormula))
}
else {
initializeModel <- PointedSDMs::startISDM(speciesDataset, Mesh = .__mesh.__, Projection = .__proj.__,
responsePA = .__responsePA.__, responseCounts = .__responseCounts.__,
trialsPA = .__trialsName.__, pointsSpatial = .__pointsSpatial.__,
pointsIntercept = .__pointsIntercept.__ , IPS = IPS,
Offset = .__Offset.__,
pointCovariates = .__pointCovariates.__,
Boundary = Workflow$.__enclos_env__$private$Area,
spatialCovariates = spatCovs,
Formulas = list(covariateFormula = Workflow$.__enclos_env__$private$covariateFormula,
biasFormula = Workflow$.__enclos_env__$private$biasFormula))
}
if (!is.null(Workflow$.__enclos_env__$private$priorsFixed)) {
for (var in names(Workflow$.__enclos_env__$private$priorsFixed)) {
initializeModel$priorsFixed(Effect = var, mean.linear = Workflow$.__enclos_env__$private$priorsFixed[[var]][1], prec.linear = Workflow$.__enclos_env__$private$priorsFixed[[var]][2])
}
}
##Here priors for the random effects
if (!is.null(Workflow$.__enclos_env__$private$copyModel)) initializeModel$specifyRandom(copyModel = .__copyModel.__)
#Workflow$specifyRandom(copyModel = x)
if (!is.null(.__pointsSpatial.__)) {
if (!is.null(Workflow$.__enclos_env__$private$sharedField)) {
if (.__pointsSpatial.__ %in% c('shared', 'correlate')) initializeModel$spatialFields$sharedField$sharedField <- Workflow$.__enclos_env__$private$sharedField
else {
for (data in names(initializeModel$spatialFields$datasetFields)) {
initializeModel$spatialFields$datasetFields[[data]] <- Workflow$.__enclos_env__$private$sharedField
}
}
}
}
if (!is.null(Workflow$.__enclos_env__$private$biasNames)) {
if (!any(names(speciesDataset) %in% Workflow$.__enclos_env__$private$biasNames)) {
warning('Bias fields specified for datasets not in the model. Turning bias off.')
biasIn <- FALSE
}
else {
biasIn <- TRUE
biasSubset <- Workflow$.__enclos_env__$private$biasNames[Workflow$.__enclos_env__$private$biasNames %in% names(speciesDataset)]
initializeModel$addBias(datasetNames = biasSubset, copyModel = Workflow$.__enclos_env__$private$biasFieldsCopy, shareModel = Workflow$.__enclos_env__$private$biasFieldsShare)
if (!is.null(Workflow$.__enclos_env__$private$biasFieldsSpecify)) {
if (any(names(speciesDataset) %in% names(Workflow$.__enclos_env__$private$biasFieldsSpecify))) {
specifySubset <- names(Workflow$.__enclos_env__$private$biasFieldsSpecify)[names(Workflow$.__enclos_env__$private$biasFieldsSpecify) %in% names(speciesDataset)]
for (biasName in specifySubset) {
initializeModel$spatialFields$biasFields[[biasName]] <- Workflow$.__enclos_env__$private$biasFieldsSpecify[[biasName]]
}
}
}
}
} else biasIn <- FALSE
if (!is.null(Workflow$.__enclos_env__$private$biasFormula)) biasIn <- TRUE
else
if (!biasIn) biasIn <- FALSE
else biasIn <- TRUE
if ('Cross-validation' %in% Oputs && 'spatialBlock' %in%Workflow$.__enclos_env__$private$CVMethod) {
initializeModel$spatialBlock(k = Workflow$.__enclos_env__$private$blockOptions$k,
rows_cols = Workflow$.__enclos_env__$private$blockOptions$rows_cols,
seed = Workflow$.__enclos_env__$private$blockOptions$seed)
}
if (initialValues) Workflow$.__enclos_env__$private$optionsINLA[['bru_initial']] <- initValues(data = initializeModel, formulaComponents = initializeModel$.__enclos_env__$private$spatcovsNames)
message('\nEstimating ISDM:\n\n')
PSDMsMOdel <- try(PointedSDMs::fitISDM(initializeModel,
options = inlaOptions))
if (inherits(PSDMsMOdel, 'try-error')) warning(paste0('Model estimation failed for ', species,'. Will skip the rest of the outputs.'))
if ('Model' %in% Oputs) {
if (saveObjects) {
if (!Quiet) message('\nSaving Model object:', '\n\n')
saveRDS(object = PSDMsMOdel, file = paste0(modDirectory,'/', speciesNameInd, '/intModel.rds'))
} else outputList[[speciesNameInd]][['Model']] <- PSDMsMOdel
}
if ('Summary' %in% Oputs) {
if ('Fixed__Effects__Comps' %in% names(PSDMsMOdel$summary.random)) {
fixedRandom <-PSDMsMOdel$summary.random$Fixed__Effects__Comps
row.names(fixedRandom) <- fixedRandom$ID
fixedRandom$ID <- NULL
} else fixedRandom <- data.frame()
if ('Bias__Effects__Comps' %in% names(PSDMsMOdel$summary.random)) {
biasFixed <-PSDMsMOdel$summary.random$Bias__Effects__Comps
row.names(biasFixed) <- biasFixed$ID
biasFixed$ID <- NULL
} else biasFixed <- data.frame()
summariesIndex <- list(Fixed = rbind(PSDMsMOdel$summary.fixed, fixedRandom, biasFixed),
Hyper = PSDMsMOdel$summary.hyperpar)
if (saveObjects) {
if (!Quiet) message('\nSaving Model summaries:', '\n\n')
saveRDS(object = summariesIndex, file = paste0(modDirectory,'/', speciesNameInd, '/modelSummary.rds'))
} else outputList[[speciesNameInd]][['Summary']] <- summariesIndex
}
if ('Cross-validation' %in% Oputs && !inherits(PSDMsMOdel, 'try-error')) {
if ('spatialBlock' %in% Workflow$.__enclos_env__$private$CVMethod) {
if (!Quiet) message('\nEstimating spatial block cross-validation:\n\n')
if (Workflow$.__enclos_env__$private$blockCVType == 'DIC') spatialBlockCV <- PointedSDMs::blockedCV(initializeModel, options = inlaOptions)
else {
blockCVPredName <- names(PSDMsMOdel$dataType)[PSDMsMOdel$dataType != 'Present only'][1]
spatialBlockCV <- PointedSDMs::blockedCV(initializeModel, options = inlaOptions, method = 'Predict', predictName = blockCVPredName)
}
if (saveObjects) {
if (!Quiet) message('\nSaving spatial blocked cross-validation object:', '\n\n')
saveRDS(object = spatialBlockCV, file = paste0(modDirectory,'/', speciesNameInd, '/spatialBlock.rds'))
} else outputList[[speciesNameInd]][['spatialBlock']] <- spatialBlockCV
rm(spatialBlockCV)
}
if ('Loo' %in% Workflow$.__enclos_env__$private$CVMethod && !inherits(PSDMsMOdel, 'try-error')) {
if (!Quiet) message('\nEstimating leave-one-out cross-validation:\n\n')
LooCV <- PointedSDMs::datasetOut(model = PSDMsMOdel)
if (saveObjects) {
if (!Quiet) message('\nSaving leave-one-out cross-validation object:', '\n\n')
saveRDS(object = LooCV, file = paste0(modDirectory,'/', speciesNameInd, '/LooCV.rds'))
} else outputList[[speciesNameInd]][['LooCV']] <- LooCV
rm(LooCV)
}
}
if (any(c('Predictions', 'Maps') %in% Oputs) && !inherits(PSDMsMOdel, 'try-error')) {
if (!Quiet) message('\nPredicting model:\n\n')
if (is.null(predictionData)) {
.__mask.__ <- as(Workflow$.__enclos_env__$private$Area, 'Spatial')
predictionData <- fmesher::fm_pixels(mesh = .__mesh.__,
mask = .__mask.__,
dims = predictionDim)
}
Predictions <- predict(PSDMsMOdel, data = predictionData, predictor = TRUE)
if (saveObjects) {
if (!Quiet) message('\nSaving predictions object:', '\n\n')
saveRDS(object = Predictions, file = paste0(modDirectory,'/', speciesNameInd, '/Predictions.rds'))
} else outputList[[speciesNameInd]][['Predictions']] <- Predictions
}
if ('Maps' %in% Oputs && !inherits(PSDMsMOdel, 'try-error')) {
if (!Quiet) message("\nPlotting Maps:\n\n")
if (saveObjects) {
if (!Quiet) message('\nSaving plots object:', '\n\n')
plot(Predictions)
ggsave(filename = paste0(modDirectory,'/', speciesNameInd, '/Map.png'))
}
else outputList[[speciesNameInd]][['Maps']] <- plot(Predictions, plot = FALSE)
rm(Predictions)
}
if ('Bias' %in% Oputs && !inherits(PSDMsMOdel, 'try-error')) {
if (biasIn) {
if (!Quiet) message('\nProducing bias predictions:\n\n')
if (is.null(predictionData)) {
.__mask.__ <- as(Workflow$.__enclos_env__$private$Area, 'Spatial')
predictionData <- fmesher::fm_pixels(mesh = .__mesh.__,
mask = .__mask.__,
dims = predictionDim)
}
biasPreds <- predict(PSDMsMOdel,
data = predictionData,
bias = TRUE)
if (saveObjects) {
if (!Quiet) message('\nSaving predictions object:', '\n\n')
saveRDS(object = biasPreds, file = paste0(modDirectory,'/', speciesNameInd, '/biasPreds.rds'))
} else outputList[[speciesNameInd]][['Bias']] <- biasPreds
}
}
}
}
}
else {
if (is.null(Workflow$.__enclos_env__$private$optionsRichness[['predictionIntercept']])) stop('predictionIntercept needs to be provided. This can be done using .$modelOptions(Richness = list(predictionIntercept = "DATASETNAME")).')
.__predIntercept.__ <- paste0(Workflow$.__enclos_env__$private$optionsRichness[['predictionIntercept']],'_intercept')
.__spatModel.__ <- Workflow$.__enclos_env__$private$optionsRichness[['speciesSpatial']]
##Fix this
#Need to get the datasets back together
spNames <- c(names(Workflow$.__enclos_env__$private$dataGBIF),
names(Workflow$.__enclos_env__$private$dataStructured))
spData <- append(unlist(Workflow$.__enclos_env__$private$dataGBIF, recursive = FALSE),
unlist(Workflow$.__enclos_env__$private$dataStructured, recursive = FALSE))
names(spData) <- spNames
wMain <- which(spNames == Workflow$.__enclos_env__$private$optionsRichness[['predictionIntercept']])
spData <- spData[c(spNames[wMain], spNames[-wMain])]
message('Setting up richness model:', '\n\n')
richSetup <- PointedSDMs::startSpecies(spData, Mesh = .__mesh.__, Projection = .__proj.__,
responsePA = .__responsePA.__, responseCounts = .__responseCounts.__,
trialsPA = .__trialsName.__, Boundary = Workflow$.__enclos_env__$private$Area,
pointsIntercept = .__pointsIntercept.__,
IPS = IPS,
pointCovariates = .__pointCovariates.__,
Offset = .__Offset.__,
speciesIntercept = Workflow$.__enclos_env__$private$speciesIntercept,
speciesName = Workflow$.__enclos_env__$private$speciesName,
speciesSpatial = .__spatModel.__,
pointsSpatial = .__pointsSpatial.__, # Make this an argument
spatialCovariates = spatCovs,
Formulas = list(covariateFormula = Workflow$.__enclos_env__$private$covariateFormula,
biasFormula = Workflow$.__enclos_env__$private$biasFormula))
##Redo this
if (!is.null(Workflow$.__enclos_env__$private$priorsFixed)) {
for (var in names(Workflow$.__enclos_env__$private$priorsFixed)) {
richSetup$priorsFixed(Effect = var, mean.linear = Workflow$.__enclos_env__$private$priorsFixed[[var]][1], prec.linear = Workflow$.__enclos_env__$private$priorsFixed[[var]][2])
}
}
if (!is.null(Workflow$.__enclos_env__$private$sharedField)) {
richSetup$spatialFields$speciesFields$speciesField <- Workflow$.__enclos_env__$private$sharedField
if (!is.null(.__pointsSpatial.__)) {
if (.__pointsSpatial.__ == 'copy') {
for (dtt in names(richSetup$spatialFields$datasetFields)) {
richSetup$spatialFields$datasetFields[[dtt]] <- Workflow$.__enclos_env__$private$sharedField
}
}
}
}
#Add copy here use specifySpatial?
if (!is.null(Workflow$.__enclos_env__$private$biasNames)) {
richSetup$addBias(datasetNames = Workflow$.__enclos_env__$private$biasNames, copyModel = Workflow$.__enclos_env__$private$biasFieldsCopy, shareModel = Workflow$.__enclos_env__$private$biasFieldsShare)
if (!is.null(Workflow$.__enclos_env__$private$biasFieldsSpecify)) {
for (biasName in names(Workflow$.__enclos_env__$private$biasFieldsSpecify)) {
richSetup$spatialFields$biasFields[[biasName]] <- Workflow$.__enclos_env__$private$biasFieldsSpecify[[biasName]]
}
}
}
#Redo this with specifyRandom
if (!is.null(Workflow$.__enclos_env__$private$priorIntercept)) richSetup$specifyRandom(speciesIntercepts = Workflow$.__enclos_env__$private$priorIntercept)
if (!is.null(Workflow$.__enclos_env__$private$optionsRichness$speciesSpatial)) {
if (!is.null(Workflow$.__enclos_env__$private$priorGroup) && Workflow$.__enclos_env__$private$optionsRichness$speciesSpatial == 'replicate') richSetup$specifyRandom(speciesGroup = Workflow$.__enclos_env__$private$priorGroup)
}
if (initialValues) Workflow$.__enclos_env__$private$optionsINLA[['bru_initial']] <- initValues(data = richSetup, formulaComponents = richSetup$.__enclos_env__$private$spatcovsNames)
message('Estimating richness model:', '\n\n')
richModel <- try(PointedSDMs::fitISDM(data = richSetup,
options = inlaOptions))
if (inherits(richModel, 'try-error')) stop('Richness model failed to estimate. Will skip the rest of the outputs.')
else
if ('Model' %in% Oputs) {
if (saveObjects) {
if (!Quiet) message('\nSaving richness model:', '\n\n')
saveRDS(object = richModel, file = paste0(modDirectory, '/richnessModel.rds')) #Add project name here
} else outputList[['RichnessModel']] <- richModel
}
if (is.null(predictionData)) {
.__mask.__ <- as(Workflow$.__enclos_env__$private$Area, 'sf')
predictionData <- fmesher::fm_pixels(mesh = .__mesh.__,
mask = .__mask.__,
dims = predictionDim)
}
if (any(c('Predictions', 'Maps') %in% Oputs) && !inherits(richModel, 'try-error')) {
richOutput <- obtainRichness(modelObject = richModel, predictionData = predictionData,
predictionIntercept = Workflow$.__enclos_env__$private$optionsRichness[['predictionIntercept']],
sampleSize = Workflow$.__enclos_env__$private$samplingSize)
if (saveObjects) {
if (!Quiet) message('\nSaving richness predictions:', '\n\n')
saveRDS(object = richOutput, file = paste0(modDirectory, '/richnessPredictions.rds')) #Add project name here
} else outputList[['Richness']] <- richOutput#richPredicts
}
#}
if ('Summary' %in% Oputs) {
removeList <- grepl('spatial', names(richModel$summary.random)) | names(richModel$summary.random) == 'speciesShared'
if (paste0(richModel$species$speciesVar,'_intercepts') %in% names(richModel$summary.random)) richModel$summary.random[[paste0(richModel$species$speciesVar,'_intercepts')]]$ID <- paste0(row.names(richModel$summary.random[[paste0(richModel$species$speciesVar,'_intercepts')]]), '_intercept')
richnessSummary <- list(Fixed = richModel$summary.fixed,
Random = do.call(rbind, richModel$summary.random[!removeList]),
Hyperparameters = richModel$summary.hyperpar)
row.names(richnessSummary$Random) <- NULL
if (saveObjects) {
if (!Quiet) message('\nSaving richness summaries:', '\n\n')
saveRDS(object = richnessSummary, file = paste0(modDirectory, '/richnessSummaries.rds')) #Add project name here
} else outputList[['Summary']] <- richnessSummary
}
if ('Bias' %in% Oputs) {
if (!Quiet) message('\nProducing bias predictions:\n\n')
if (is.null(predictionData)) {
.__mask.__ <- as(Workflow$.__enclos_env__$private$Area, 'Spatial')
predictionData <- fmesher::fm_pixels(mesh = .__mesh.__,
mask = .__mask.__,
dims = predictionDim)
}
biasPreds <- predict(richModel,
data = predictionData,
bias = TRUE)
if (saveObjects) {
if (!Quiet) message('\nSaving predictions object:', '\n\n')
saveRDS(object = biasPreds, file = paste0(modDirectory,'/', '/biasRichnessPreds.rds')) #Add project name here
} else outputList[['BiasRichness']] <- biasPreds
}
}
if (!saveObjects) {
if (!exists('outputList')) stop('Workflow did not run for any species.')
return(outputList)
}
}
Try the intSDM package in your browser
Any scripts or data that you put into this service are public.
R Package Documentation
Browse R Packages
We want your feedback!
Note that we can't provide technical support on individual packages. You should contact the package authors for that.
Embedding an R snippet on your website
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.