tests/testthat/test_part_partitionOccs.R

In wallace: A Modular Platform for Reproducible Modeling of Species Niches and Distributions

#### COMPONENT part: Partition Occurrence Data
#### MODULE: Non-spatial Partition & Spatial Partition
context("partitionOccs")

### Set parameters

## occurrences
occs <- read.csv(system.file("extdata/Bassaricyon_alleni.csv",
                 package = "wallace"))
bgMask <- envs_userEnvs(rasPath = list.files(system.file("extdata/wc",
                                             package = "wallace"),
                        pattern = ".tif$", full.names = TRUE),
                        rasName = list.files(system.file("extdata/wc",
                                             package = "wallace"),
                        pattern = ".tif$", full.names = FALSE))
bgSample <- read.csv(system.file("extdata/Bassaricyon_alleni_bgPoints.csv",
                                 package = "wallace"))


# partition methodology
jack <- 'jack' # Non-spatial Partition - jackknife
folds <- 'rand' # Non-spatial Partition - random k-fold
block <- 'block' # spatial Partition - block
cb1 <- 'cb1' # spatial Partition - checkerboard 1 (K=2)
cb2 <- 'cb2' # spatial Partition - checkerboard 2 (K=4)

## number of folds to partition data
kfolds <- 4

# aggregation factor
aggFact <- 2

### run function

## jackknife
partJack <- part_partitionOccs(occs, bgSample, method = jack, kfolds = NULL,
                               bgMask = NULL, aggFact = NULL)
## random k-fold
partfold <- part_partitionOccs(occs, bgSample, method = folds, kfolds,
                               bgMask = NULL, aggFact = NULL)
## block
partblock <- part_partitionOccs(occs, bgSample, method = block, kfolds = NULL,
                                bgMask = NULL, aggFact = NULL)
## checkerboard 1
partche1 <- part_partitionOccs(occs, bgSample, method = cb1, kfolds = NULL,
                               bgMask, aggFact)
## checkerboard 2
partche2 <- part_partitionOccs(occs, bgSample, method = cb2, kfolds = NULL,
                               bgMask, aggFact)


## test if the error messages appear when they are supposed to
test_that("error checks", {
  # the user has not selected a partitioning option
  expect_error(part_partitionOccs(occs, bgSample, method = '', kfolds = NULL,
                                  bgMask = NULL, aggFact = NULL),
               'Please select a partitioning option.')
  # method is random partition and kfolds parameter is NULL
  expect_error(part_partitionOccs(occs, bgSample, method = folds, kfolds = NULL,
                                  bgMask = NULL, aggFact = NULL),
               "Please specify a kfold value to use the random partition function.",
               fixed = TRUE)
  # method is random partition and kfolds < 2
  expect_error(part_partitionOccs(occs, bgSample, method = folds, kfolds = 1,
                                  bgMask = NULL, aggFact = NULL),
               "Please specify a kfold value greater than 1.", fixed = TRUE)
  # method is checkerboard 1 but aggregation factor parameter < 1
  expect_error(part_partitionOccs(occs, bgSample, method = cb1, kfolds = NULL,
                                  bgMask, aggFact = -2),
               "Please specify a positive aggregation factor greater than 1.",
               fixed = TRUE)
  # method is checkerboard 2 but aggregation factor parameter < 1
  expect_error(part_partitionOccs(occs, bgSample, method = cb2, kfolds = NULL,
                                  bgMask, aggFact = -2),
               "Please specify a positive aggregation factor greater than 1.",
               fixed = TRUE)
  # method is checkerboard 1 but aggregation factor parameter = 1
  expect_error(part_partitionOccs(occs, bgSample, method = cb1, kfolds = NULL,
                                  bgMask, aggFact = 1),
               "Please specify a positive aggregation factor greater than 1.",
               fixed = TRUE)
  # method is checkerboard 2 but aggregation factor parameter = 1
  expect_error(part_partitionOccs(occs, bgSample, method = cb2, kfolds = NULL,
                                  bgMask, aggFact = 1),
               "Please specify a positive aggregation factor greater than 1.",
               fixed = TRUE)
  # method is checkerboard 1 but aggregation factor parameter is NA
  expect_error(part_partitionOccs(occs, bgSample, method = cb1, kfolds = NULL,
                                  bgMask, aggFact = NA),
               "Please specify a positive aggregation factor greater than 1.",
               fixed = TRUE)
  # method is checkerboard 2 but aggregation factor parameter is NA
  expect_error(part_partitionOccs(occs, bgSample, method = cb2, kfolds = NULL,
                                  bgMask, aggFact = NA),
               "Please specify a positive aggregation factor greater than 1.",
               fixed = TRUE)
  # method is checkerboard 1 but aggregation factor parameter is NULL
  expect_error(part_partitionOccs(occs, bgSample, method = cb1, kfolds = NULL,
                                  bgMask, aggFact = NULL),
               "Please specify an aggregation factor to use checkerboard partition functions.",
               fixed = TRUE)
  # method is checkerboard 2 but aggregation factor parameter is NULL
  expect_error(part_partitionOccs(occs, bgSample, method = cb2, kfolds = NULL,
                                  bgMask, aggFact = NULL),
               "Please specify an aggregation factor to use checkerboard partition functions.",
               fixed = TRUE)
  # method is checkerboard 1 but athe bgMask parameter is NULL
  expect_error(part_partitionOccs(occs, bgSample, method = cb1, kfolds = NULL,
                                  bgMask = NULL, aggFact = 2),
               "Please specify a background mask to use checkerboard partition functions.",
               fixed = TRUE)
  # method is checkerboard 2 but athe bgMask parameter is NULL
  expect_error(part_partitionOccs(occs, bgSample, method = cb2, kfolds = NULL,
                                  bgMask = NULL, aggFact = 2),
               "Please specify a background mask to use checkerboard partition functions.",
               fixed = TRUE)
})

### test output features
test_that("output checks", {
  # the output is a list
  expect_is(partJack, "list")
  expect_is(partfold, "list")
  expect_is(partblock, "list")
  expect_is(partche1, "list")
  expect_is(partche1, "list")
  # the list has two elements
  expect_equal(length(partJack), 2)
  expect_equal(length(partfold), 2)
  expect_equal(length(partblock), 2)
  expect_equal(length(partche1), 2)
  expect_equal(length(partche2), 2)
  # jackknife: the number of partition is the same as the number of records
  expect_equal(nlevels(factor(partJack$occs.grp)), nrow(occs))
  # random: the number of partition is the same as specified in the function (kfolds).
  expect_equal(nlevels(factor(partfold$occs.grp)), kfolds)
  # block: the number of partition is 4
  expect_equal(nlevels(factor(partblock$occs.grp)), 4)
  # checkerboard 1: the number of partition is 2
  expect_equal(nlevels(factor(partche1$occs.grp)), 2)
  # checkerboard 2: the number of partition is 4
  expect_equal(nlevels(factor(partche2$occs.grp)), 4)
})