tests/testthat/test_kmedians_cluster.R

In UBC-MDS/KMediansR: K-Medians Clustering

#' kmedians_cluster tests
#'
#' Testing includes:
#'  checks errors of the missing input
#'  checks errors of the datatype of the input
#'  checks if num_clusters is larger than the data rows
#'  checks the datatype of the output
#'  checks the size of the output
#'  correct medians and clustering of the toy example

context("testing kmedians clustering")


# toy data for calculation
A <- matrix(
  c(1,1,1,2,2,1,100,100,101,100,100,101),
  nrow = 6,
  ncol = 2,
  byrow = TRUE)

n <- 2

test_that("missing input errors", {

  expect_error(kmedians(X = NULL, num_clusters = n),

               'Input X should be a matrix!')

  expect_error(kmedians(X = A, num_clusters = NULL),

               'non-numeric argument to mathematical function')

})


test_that("datatype errors", {

  expect_error(kmedians(X = "abcd", num_clusters = n),

               'Input X should be a matrix!')

  expect_error(kmedians(X = 100, num_clusters = n),

               'Input X should be a matrix!')

  expect_error(kmedians(X = c(1,2), num_clusters = n),

               'Input X should be a matrix!')

  expect_error(kmedians(X = A, num_clusters = 1.5),

               'Input number of clusters should be an integer!')

  expect_error(kmedians(X = A, num_clusters = "2"),

               'non-numeric argument to mathematical function')

  expect_error(kmedians(X = A, num_clusters = 0),

               'subscript out of bounds')

})

test_that("error if num_clusters is larger than the data rows", {

  expect_error(kmedians(X = A, num_clusters = nrow(A)+1),

               "cannot take a sample larger than the population when 'replace = FALSE'")

})


test_that("test if output is a list given toy data", {

  expect_equal(is.list(kmedians(X = A, num_clusters = n)), TRUE)

})


test_that("test size of output given toy data", {

  expect_equal(length(kmedians(X = A, num_clusters = n)), 2)

  expect_equal(nrow(kmedians(X = A, num_clusters = n)[[1]]), 2)
  # number of medians equals to number of clusters

  expect_equal(length(kmedians(X = A, num_clusters = n)[[2]]), 6)
  # number of labels equals to number of data points
})

test_that("test if the medians and clustering given toy data are correct", {

  expect_equal(kmedians(X = A, num_clusters = n)[[1]][1,1],1)

  expect_equal(kmedians(X = A, num_clusters = n)[[1]][2,1],100)

  expect_equal(length(unique(kmedians(X = A, num_clusters = n)[[2]])), 2)

  expect_equal(kmedians(X = A, num_clusters = n)[[2]],c(1,1,1,2,2,2))

})