cleaned habitat modeling scripts

analysis/habitat_model/main_habitat.R

@@ -0,0 +1,82 @@
+#---------------------------------------------------------------
+# main_habitat_species
+#---------------------------------------------------------------
+# This script controls the workflow used for habitat modeling
+
+
+#-----------------------------------------------------------
+# 1. Set parameters
+#-----------------------------------------------------------
+
+## Load dependencies
+source("setup.R")
+source("scr/fun_habitat_plot.R")
+
+## Set initial parameters
+sp_code <- "GAZ"  # species code
+sp_name <- "A. gazella"
+stack_repo <- paste0(output_data, "/stack_daily")  # location of environmental data
+vars <- c("BAT", "SLP", "SST", "SSTg", "SAL", "SALg", "SSH", "EKE", "CHL", "SIC", "MLD", "EDGE", "SIT")  # list of all predictors
+
+## computer cores
+cores <- 50  # numbers of cores
+
+
+#-----------------------------------------------------------
+# 2. Split training and testing dataset
+#-----------------------------------------------------------
+# Training and testing
+# This step was originaly designed to split datasets for cross-validation.
+# Final version of the analysis uses a leave-one-out approach and this part is
+# is no longer required. I keep the code as it does not affect and can be useful
+# for other works.
+train_prop <- 0.75
+source("analysis/habitat_model/scr/training_testing.R")
+
+
+#-----------------------------------------------------------
+# 3. Exploratory Data Analysis
+#-----------------------------------------------------------
+# Reads all presence/absence data and generates several plots to explore all variables
+# - Checks for missing data
+# - Pearson and Spearman rank correlations
+# - Density plots to compare presence/absence
+source("analysis/habitat_model/scr/eda.R")
+
+# After visual inspection, keep and exclude variables
+vars2remove <- c("SIT")  # We remove SIT (keep SIC) 
+vars <- vars[!vars %in% vars2remove]
+
+
+#-----------------------------------------------------------
+# 4. Fit BRT
+#-----------------------------------------------------------
+# Fit model
+source("analysis/habitat_model/scr/fit_brt.R")
+
+
+
+#-----------------------------------------------------------
+# 5. Fit Accessibility model
+#-----------------------------------------------------------
+# Fit model
+source("analysis/habitat_model/scr/accessibility_ice.R")
+
+
+#-----------------------------------------------------------
+# 6. Predict combining BRT and accessbility model
+#-----------------------------------------------------------
+# Set prediction period
+date_start <- as.Date("2019-02-01")
+date_end <- as.Date("2019-09-30")
+
+# Select if use bootrap models or full model
+bootstrap <- TRUE
+
+# Predict
+if(bootstrap == FALSE) source("analysis/habitat_model/scr/predict_brt.R")
+if(bootstrap == TRUE) source("analysis/habitat_model/scr/predict_brt_boot.R")
+
+
+
+