inst/tinytest/test-kplus-criterion.R

In anticlust: Subset Partitioning via Anticlustering

library("anticlust")

# Test that k-plus criterion is computed correctly
set.seed(123)

init <- anticlust:::initialize_clusters(96, 3, NULL)

df <- schaper2019[, 3:6]

groups1 <- anticlustering(
  df,
  K = init,
  objective = "kplus"
)

groups2 <- anticlustering(
  cbind(df, anticlust:::squared_from_mean(df)),
  K = init,
  objective = "variance"
)

groups3 <- anticlustering(
  kplus_moment_variables(df, T = 2, standardize = FALSE), 
  K = init,
  objective = "variance"
)

expect_true(all(groups1 == groups2))
expect_true(all(groups1 == groups3))

# add test with standardize argument
groups4 <- anticlustering(
  df,
  K = init,
  objective = "kplus",
  standardize = TRUE
)

groups5 <- anticlustering(
  scale(cbind(df, anticlust:::squared_from_mean(df))),
  K = init,
  objective = "variance",
  standardize = FALSE
)

groups6 <- anticlustering(
  kplus_moment_variables(df, T = 2), 
  K = init,
  objective = "variance"
)

expect_true(all(groups4 == groups5))
expect_true(all(groups4 == groups6))