In biomodhub/biomod2: Ensemble Platform for Species Distribution Modeling

knitr::opts_chunk$set(echo = TRUE, comment = NA)
options(width = 120)
library(biomod2)
library(terra)
library(data.table)

bm.calib = NULL
hasEvalMod = hasEvalEns = FALSE
hasBmFiles = !is.na(bm.files)
hasBmMod = !is.na(bm.mod)
hasBmEns = !is.na(bm.ens)
hasBmForm = !is.na(bm.form)
color1 = params.color$color1
color2 = params.color$color2
color3 = params.color$color3

bm.calib = get_calib_lines(bm.mod)

name.all = colnames(bm.calib)
name.PA = sapply(name.all, function(x) strsplit(x, "_")[[1]][2])
name.PA = unique(name.PA)
name.run = sapply(name.all, function(x) strsplit(x, "_")[[1]][3])
name.run = unique(name.run)
nb.PA = sum(!grepl("allData", name.PA))
nb.rep = sum(!grepl("allRun", name.run))

hasPA = inherits(bm.form, "BIOMOD.formated.data.PA")
hasDataEval = ifelse("has.data.eval" %in% slotNames(bm.form), bm.form@has.data.eval, FALSE)
hasFilterRaster = ifelse("has.filter.raster" %in% slotNames(bm.form)
                         , bm.form@has.filter.raster, FALSE) ##

# form.summ = summary(bm.form, calib.lines = bm.calib) ## TODO CORRECT
form.summ = data.frame("dataset" = "initial",
                       "run" = NA,
                       "PA" = NA,
                       "Presences" = length(which(bm.form@data.species == 1)),
                       "True_Absences" = length(which(bm.form@data.species == 0)),
                       "Pseudo_Absences" = 0,
                       "Undefined" = length(which(is.na(bm.form@data.species))))

nb.1 = form.summ$Presences[which(form.summ$dataset == "initial")]
nb.0 = form.summ$True_Absences[which(form.summ$dataset == "initial")]
nb.NA = form.summ$Pseudo_Absences[which(form.summ$dataset == "calibration" &
                                          form.summ$run == "allRun" &
                                          form.summ$PA == "PA1")]

CV.perc = round(100 * form.summ$Presences[which(form.summ$dataset == "calibration" &
                                                  form.summ$run == "RUN1" &
                                                  form.summ$PA == "allData")] / nb.1) ##
CV.perc.PA = round(100 * form.summ$Presences[which(form.summ$dataset == "calibration" &
                                                  form.summ$run == "RUN1" &
                                                  form.summ$PA == "PA1")] / nb.1) ##

name.var = colnames(bm.form@data.env.var)

eval.mod = get_evaluations(bm.mod)
hasEvalMod = !is.null(eval.mod)

eval.ens = get_evaluations(bm.ens)
hasEvalEns = !is.null(eval.ens)

# name.var = get_formal_data(bm.mod, subinfo = "expl.var.names")
name.algo = get_built_models(bm.mod)
name.algo = sapply(name.algo, function(x) strsplit(x, "_")[[1]][4])
name.algo = sort(unique(name.algo))

name.ens = get_built_models(bm.ens)
name.ens = sapply(name.ens, function(x) strsplit(x, "_")[[1]][2])
name.ens = sort(unique(name.ens))

name.eval.mod = name.eval.ens = NULL
evalSame = FALSE
if (hasEvalMod) {
  name.eval.mod = sort(unique(eval.mod$metric.eval))
}
if (hasEvalEns) {
  name.eval.ens = sort(unique(eval.ens$metric.eval))
}
if (hasEvalMod && hasEvalEns) {
  evalSame = (length(name.eval.mod) == length(name.eval.ens) &&
              all.equal(name.eval.mod, name.eval.ens))
}

# mod.list = get_built_models(bm.mod)
# i = 12
# BIOMOD_LoadModels(bm.mod, full.name = mod.list[i])
# mod.i = get(mod.list[i])
# hop = get_formal_model(mod.i)
# # if GAM, GLM, MARS hop$coefficients
# rm(list = mod.list[i])

ODMAP protocol

data(ODMAP)

fun_ind <- function(section, subsection, element) {
  return(which(ODMAP$section == section &
                    ODMAP$subsection == subsection &
                    ODMAP$element == element))
}

## Information to be filled by hand by user
ODMAP$value[fun_ind("Overview", "Authorship", "Study title")] = "tobefilledbyuser"
ODMAP$value[fun_ind("Overview", "Authorship", "Author names")] = "tobefilledbyuser"
ODMAP$value[fun_ind("Overview", "Authorship", "Contact")] = "tobefilledbyuser"
ODMAP$value[fun_ind("Overview", "Authorship", "Study link")] = "tobefilledbyuser"
ODMAP$value[fun_ind("Overview", "Model objective", "Target output")] = "tobefilledbyuser"
ODMAP$value[fun_ind("Overview", "Location", "Location")] = "tobefilledbyuser"
ODMAP$value[fun_ind("Overview", "Hypotheses", "Hypotheses")] = "tobefilledbyuser"
ODMAP$value[fun_ind("Overview", "Assumptions", "Model assumptions")] = "tobefilledbyuser"

## Information to be chosen by user
ODMAP$value[fun_ind("Overview", "Model objective", "Model objective")] = params.ODMAP$O.mod.objective
ODMAP$value[fun_ind("Overview", "Scale of Analysis", "Boundary")] = params.ODMAP$O.boundary
ODMAP$value[fun_ind("Overview", "Biodiversity data", "Observation type")] = paste0(params.ODMAP$O.obs.type, collapse = ", ")
ODMAP$value[fun_ind("Overview", "Predictors", "Predictor types")] = paste0(params.ODMAP$O.pred.type, collapse = ", ")
ODMAP$value[fun_ind("Data", "Biodiversity data", "Ecological level")] = params.ODMAP$D.eco.level
ODMAP$value[fun_ind("Data", "Biodiversity data", "Sampling design")] = params.ODMAP$D.samp.design

## Species name
ODMAP$value[fun_ind("Overview", "Focal Taxon", "Focal Taxon")] = sp.name
ODMAP$value[fun_ind("Data", "Biodiversity data", "Taxon names")] = sp.name

if (!is.null(bm.form@data.mask)) {
  mm = unwrap(bm.form@data.mask$calibration)

  ## Overview
  ODMAP$value[fun_ind("Overview", "Scale of Analysis", "Spatial extent")] = as.character(ext(mm))
  ODMAP$value[fun_ind("Overview", "Scale of Analysis", "Spatial resolution")] = paste0(res(mm), collapse = ", ")

  ## Predictor variables
  ODMAP$value[fun_ind("Data", "Predictor variables", "Spatial extent")] = as.character(ext(mm))
  ODMAP$value[fun_ind("Data", "Predictor variables", "Spatial resolution")] = paste0(res(mm), collapse = ", ")
  ODMAP$value[fun_ind("Data", "Predictor variables", "Coordinate reference system")] = crs(mm, proj = TRUE)
}

ODMAP$value[fun_ind("Overview", "Biodiversity data", "Response data type")] = ifelse(hasPA, "presence-only", "presence-absence")

if (hasBmMod) {
  ODMAP$value[fun_ind("Overview", "Algorithms", "Modelling techniques")] = paste0(name.algo, collapse = ", ")
}
if (hasBmEns) {
  ODMAP$value[fun_ind("Overview", "Algorithms", "Model averaging")] = paste0(name.ens, collapse = ", ")
}
# ODMAP$value[fun_ind("Overview", "Software", "Software")] = paste0("biomod2 v", packageVersion("biomod2"))
ODMAP$value[fun_ind("Overview", "Software", "Software")] = paste0("biomod2 v", bm.form@biomod2.version)


if (hasFilterRaster){
  ODMAP$value[fun_ind("Data", "Biodiversity data", "Scaling")] = "Thinning to keep only one occurrence per cell"
} else {
  ODMAP$value[fun_ind("Data", "Biodiversity data", "Scaling")] = "none"
}

if (hasPA) {
  ODMAP$value[fun_ind("Data", "Biodiversity data", "Absence data")] = "none"
  ODMAP$value[fun_ind("Data", "Biodiversity data", "Background data")] = paste(nb.PA, "sets of", nb.NA, "pseudo-absences selected with", bm.form@PA.strategy, "strategy")
} else {
  ODMAP$value[fun_ind("Data", "Biodiversity data", "Absence data")] = paste(nb.0, "true absences")
  ODMAP$value[fun_ind("Data", "Biodiversity data", "Background data")] = "none"
}

ODMAP$value[fun_ind("Data", "Biodiversity data", "Sample size")] = paste0(c(paste(nb.1, "presences")
                                                                , ifelse(nb.0 > 0, paste(nb.0, "true absences"), "")), collapse = ", ")

if (hasBmMod) {
if (hasPA) {
    ODMAP$value[fun_ind("Data", "Data partitioning", "Training data")] = ifelse(nb.rep == 0, "all", paste(nb.rep, "sets of", CV.perc.PA, "percent of the data for each pseudo-absence dataset"))
  ODMAP$value[fun_ind("Data", "Data partitioning", "Validation data")] = ifelse(nb.rep == 0, "none", paste(nb.rep, "sets of", 100 - CV.perc.PA, "percent of the data for each pseudo-absence dataset"))
} else {
  ODMAP$value[fun_ind("Data", "Data partitioning", "Training data")] = ifelse(nb.rep == 0, "all", paste(nb.rep, "sets of", CV.perc, "percent of the data"))
  ODMAP$value[fun_ind("Data", "Data partitioning", "Validation data")] = ifelse(nb.rep == 0, "none", paste(nb.rep, "sets of", 100 - CV.perc, "percent of the data"))
}
}
ODMAP$value[fun_ind("Data", "Data partitioning", "Test data")] = ifelse(hasDataEval, "TRUC", "none")

ODMAP$value[fun_ind("Data", "Predictor variables", "Predictor variables")] = paste0(name.var, collapse = ", ")

ODMAP$value[fun_ind("Model", "Model estimates", "Coefficient")] = paste0(name.var, collapse = ", ")

if (hasEvalMod || hasEvalEns) {
  if (evalSame) {
    mess = paste0(paste0(name.eval.mod, collapse = ", "), " for all models")
  } else {
    mess = paste0(paste0(name.eval.mod, collapse = ", "), " for single models ; "
                  , paste0(name.eval.ens, collapse = ", "), " for ensemble models")
  }
} else { mess = NA }
ODMAP$value[fun_ind("Assessment", "Performance statistics", "Performance on training data")] = mess
ODMAP$value[fun_ind("Assessment", "Performance statistics", "Performance on validation data")] = mess
ODMAP$value[fun_ind("Assessment", "Performance statistics", "Performance on test data")] = mess
if (hasBmMod) {
  ODMAP$value[fun_ind("Assessment", "Plausibility check", "Response shapes")] = "Response curves"
}

fwrite(ODMAP, file = paste0("biomod2_ODMAP_", sp.name, "_table.csv")
       , col.names = FALSE, row.names = FALSE, quote = FALSE, sep = "\t")

fun_print <- function(tmp, rm.na = FALSE) {
  if (rm.na) {
    tmp = tmp[which(!is.na(tmp$value)), ]
  }
  subsection.prev = text.toprint = ""
  for(i in 1:nrow(tmp)){
    if (tmp$subsection[i] == subsection.prev) {
      cat.subsection = ""
    } else {
      cat.subsection = paste0(ifelse(i == 1, "", "\n"), "\t", tmp$subsection[i])
      subsection.prev = tmp$subsection[i]
    }
    cat.element = tmp$element[i]
    cat.value = tmp$value[i]
    nb.tab = ifelse(nchar(cat.element) < 8, 6
                    , ifelse(nchar(cat.element) < 12, 5
                             , ifelse(nchar(cat.element) < 16, 4
                                      , ifelse(nchar(cat.element) < 20, 3, 2))))
    text.toprint = c(text.toprint, paste0(cat.subsection
                                        , "\n\t\t"
                                        , cat.element
                                        , paste0(rep("\t", nb.tab), collapse = "")
                                        , cat.value))
  }
  text.toprint = paste0(text.toprint, collapse = "")
  # text.toprint = gsub("tobefilledbyuser", "*tobefilledbyuser*", text.toprint)
  return(text.toprint)
}

Overview

tmp = ODMAP[which(ODMAP$section == "Overview"), ]
# cat(fun_print(tmp, rm.na = TRUE))
cat(fun_print(tmp, rm.na = FALSE))

Data

tmp = ODMAP[which(ODMAP$section == "Data"), ]
if (any(!is.na(tmp$value))) {
  cat(fun_print(tmp, rm.na = FALSE))
}

Model

tmp = ODMAP[which(ODMAP$section == "Model"), ]
if (any(!is.na(tmp$value))) {
  cat(fun_print(tmp, rm.na = FALSE))
}

Assessment

tmp = ODMAP[which(ODMAP$section == "Assessment"), ]
if (any(!is.na(tmp$value))) {
  cat(fun_print(tmp, rm.na = FALSE))
}

Prediction

tmp = ODMAP[which(ODMAP$section == "Prediction"), ]
if (any(!is.na(tmp$value))) {
  cat(fun_print(tmp, rm.na = FALSE))
}

biomodhub/biomod2 documentation built on March 29, 2025, 1:14 p.m.

rdrr.io home R language documentation Run R code online

CRAN packages Bioconductor packages R-Forge packages GitHub packages

Note that we can't provide technical support on individual packages. You should contact the package authors for that.

biomodhub/biomod2
Ensemble Platform for Species Distribution Modeling

In biomodhub/biomod2: Ensemble Platform for Species Distribution Modeling

ODMAP protocol

Overview

Data

Model

Assessment

Prediction

R Package Documentation

Browse R Packages

We want your feedback!

biomodhub/biomod2 Ensemble Platform for Species Distribution Modeling

In biomodhub/biomod2: Ensemble Platform for Species Distribution Modeling

ODMAP protocol

Overview

Data

Model

Assessment

Prediction

R Package Documentation

Browse R Packages

We want your feedback!

biomodhub/biomod2
Ensemble Platform for Species Distribution Modeling