tests/testthat/test-predict.R

In UBC-MDS/KmeansR: Kmeans Algorithm Implemented in R

library(testthat)
library(tidyverse)

context("predict")

tests <- function() {
  #helper data
  data_new <- data.frame("x1" = c(1, 1,  15, 1,  3),
                         "x2" = c(2, 6,  2,  10, 10),
                         "x3" = c(3, 10, 6,  11, 11))
  data_new2 <- data.frame("x1" = c(1, 1,  15, 1,  3, 10, 11, 4, 2),
                          "x2" = c(2, 6,  2,  10, 10, 1, 43, 4, 1),
                          "x3" = c(3, 10, 6,  11, 11, 3, 10, 3, 20))
  data_new3 <- data.frame("x1" = c(1, 1,  15, 1,  3),
                         "x2" = c(2, 6,  2,  10, 10),
                         "x3" = c("a", 10, 6,  11, 11))
  centroids <- list(c(1,2,3), c(1,6,10), c(15,2,6), c(1,10,11), c(3,10,11))
  centroids2 <- list(c(1,2,3), c(1,6,10), c(15,2,6), c(1,10,11), c(3,10,11,12))
  centroids3 <- list(c("a",2,3), c(1,6,10), c(15,2,6), c(1,10,11), c(3,10,11))
  centroids4 <- data.frame(c(1,2,3,4,5), c(1,2,3,6,10), c(15,1,3,2,6))
  
  #tests
  test_that("Each points in new data needs to be assigned a centroid", {
    expect_equal(length(predict(data_new2, centroids)), nrow(data_new2))
  })
  
  test_that("Points with same coordinates as centroids should be assigned to that centroid", {
    expect_equal(predict(data_new, centroids), c(1, 2, 3, 4, 5))
  })
  
  test_that("Dimensions of centroids must be the same as dimension of data", {
    expect_error(predict(data_new, centroids2))
  })
  
  test_that("Data must be numeric", {
    expect_error(predict(data_new3, centroids))
  })
  
  test_that("Centroid coordinates must be numeric", {
    expect_error(predict(data_new, centroids3))
  })
  
  test_that("Centroid may be a dataframe", {
    expect_equal(length(predict(data_new2, centroids4)), nrow(data_new2))
  })
  
}

tests()