Predict from SDM — predict

The resulting pred.tif is masked to the boundary provided to the pred_limit argument of prep_sdm; or generated in prep_sdm from the pred_limit, limit_buffer and pred_clip arguments.

Usage

predict_sdm(
  prep,
  full_run,
  out_dir,
  file_name = "pred.tif",
  use_env_naming = FALSE,
  predictors = NULL,
  is_env_pred = FALSE,
  terra_options = NULL,
  doClamp = TRUE,
  force_new = FALSE,
  do_gc = FALSE,
  check_tifs = FALSE,
  handle_errors = TRUE,
  ...
)

Arguments

prep: Character or named list. If character, the path to an existing prep.rds. Otherwise, the result of a call to prep_sdm() with return_val = "object".
full_run: Character or named list. If character, the path to an existing full_run.rds. Otherwise, the result of a call to run_full_sdm() with return_val = "object".
out_dir: Character. Name of directory into which .tifs will be saved. Will be created if it does not exist.
file_name: Character. Name to give the output prediction .tif.
use_env_naming: Logical. If TRUE, and is_env_pred is TRUE, naming will ignore file_name and instead generate a name matching name_env_tif() with layer being this_taxa from prep and start_date being the minimum available start_date from the predictors. pred appears between this_taxa and start_date.
predictors: Character. Vector of paths to predictor .tif files.
is_env_pred: Logical. Does the naming of the directory and files in predictors follow the pattern required by envRaster::parse_env_tif()?
terra_options: Passed to terra::terraOptions(). e.g. list(memfrac = 0.6)
doClamp: Passed to terra::predict() (which then passes as ... to fun). Possibly orphaned from older envSDM?
force_new: Logical. If output files already exist, should they be remade?
do_gc: Logical. Run base::rm(list = ls) and base::gc() at end of function? Useful to keep RAM use down when running SDMs for many, many taxa, especially if done in parallel.
check_tifs: Logical. Check if any output .tif files error on terra::rast() and delete them if they do. Useful after a crash during predict.
handle_errors: Logical. Use purrr::safely when predicting, enabling the capture of (m)any errors (which are then written to the log). Suggest turning off (i.e. handle_errors = FALSE) when running in a targets pipeline.
...: Passed to ... in terra::mask() - the last step in the envSDM::predict_sdm process. Used to provide additional arguments to terra::writeRaster.

Value

Character path to predicted file, usually 'pred.tif'. Output .tif and .log, written to out_dir.

Examples


  # setup -------
  out_dir <- file.path(system.file(package = "envSDM"), "examples")

  # predictors -------
  preds <- fs::dir_ls(fs::path(out_dir, "tif"))


  # data ------
  data <- fs::path(system.file(package = "envSDM"), "examples") |>
    fs::dir_ls(regexp = "prep\\.rds$"
               , recurse = TRUE
               ) |>
    tibble::enframe(name = NULL, value = "prep") |>
    dplyr::mutate(taxa = gsub("\\.rds", "", basename(dirname(prep)))
                  , tune = gsub("prep", "tune", prep)
                  , out_dir = fs::path(out_dir, taxa, "combo")
                  )

  # Best combo--------
  ## run full SDM --------
  purrr::pmap(list(data$prep
                    , data$tune
                    , data$out_dir
                    )
               , \(a, b, c) run_full_sdm(prep = a
                                         , tune = b
                                         , out_dir = c
                                         , use_metric = "combo"

                                         # passed to tune_sdm via dots
                                         , metrics_df = envSDM::sdm_metrics
                                         , force_new = FALSE
                                         )
               )
#> [[1]]
#> C:/temp/nige/RtmpkDfFu6/temp_libpath57ecc1774b9/envSDM/examples/chg/combo/full_run.rds
#> 
#> [[2]]
#> C:/temp/nige/RtmpkDfFu6/temp_libpath57ecc1774b9/envSDM/examples/mjs/combo/full_run.rds
#> 
#> [[3]]
#> C:/temp/nige/RtmpkDfFu6/temp_libpath57ecc1774b9/envSDM/examples/wjb/combo/full_run.rds
#> 


  ## predict -------
  purrr::pwalk(list(data$prep
                    , data$out_dir
                    )
               , \(a, b) predict_sdm(prep = a
                                     , full_run = fs::path(b, "full_run.rds")
                                     , out_dir = b
                                     , predictors = preds
                                     , check_tifs = TRUE
                                     , force_new = FALSE
                                     )
               )

  ## .pngs -------
  if(FALSE) {

    # not working for binary 'thresh' rasters (due to an issue with number of 'classes"?):
      # Error in if (any(na.omit(x) < min(breaks)) && show.warnings) warning("Values have found that are less than the lowest break",  :
      # missing value where TRUE/FALSE needed
    purrr::walk2(data$out_dir
                 , data$out_dir
                 , \(x, y) png_from_preds(pred_dir = x
                                          , tune_dir = y
                                          , trim = FALSE
                                          , recurse = 1
                                          )
                 )

  }

  ## visualise-------
  ### mask -------
  purrr::walk(data$out_dir[file.exists(fs::path(data$out_dir, "pred.tif"))]
              , \(x) fs::path(x, "pred.tif") %>%
                terra::rast() %>%
                terra::trim() %>%
                terra::plot()
              )



  # Best auc--------
  ## run full SDM --------
  data <- data %>%
    dplyr::mutate(out_dir = gsub("combo", "auc_po", out_dir))

  purrr::pmap(list(data$prep
                    , data$tune
                    , data$out_dir
                    )
               , \(a, b, c) run_full_sdm(prep = a
                                         , tune = b
                                         , out_dir = c
                                         , use_metric = "auc_po"

                                         # passed to tune_sdm via dots
                                         , metrics_df = envSDM::sdm_metrics
                                         , force_new = FALSE
                                         )
              )
#> [[1]]
#> C:/temp/nige/RtmpkDfFu6/temp_libpath57ecc1774b9/envSDM/examples/chg/auc_po/full_run.rds
#> 
#> [[2]]
#> C:/temp/nige/RtmpkDfFu6/temp_libpath57ecc1774b9/envSDM/examples/mjs/auc_po/full_run.rds
#> 
#> [[3]]
#> C:/temp/nige/RtmpkDfFu6/temp_libpath57ecc1774b9/envSDM/examples/wjb/auc_po/full_run.rds
#> 

  ## predict -------
  purrr::pwalk(list(data$prep
                    , data$out_dir
                    )
               , \(a, b) predict_sdm(prep = a
                                     , full_run = fs::path(b, "full_run.rds")
                                     , out_dir = b
                                     , predictors = preds
                                     , is_env_pred = FALSE
                                     , check_tifs = TRUE
                                     , force_new = FALSE
                                     )
               )


  ## visualise-------
  ### mask -------
  purrr::walk(data$out_dir[file.exists(fs::path(data$out_dir, "pred.tif"))]
              , \(x) fs::path(x, "pred.tif") %>%
                terra::rast() %>%
                terra::trim() %>%
                terra::plot()
              )