A test for testing the null hypothesis of random agreement (i.e., adjusted Rand Index equal to 0) between two partitions.

A test for testing the null hypothesis of random agreement (i.e., adjusted Rand Index equal to 0) between two partitions.

cc_test_ari(ground_truth, partition)

Arguments

ground_truth

(int) A vector of the actual membership of elements in clusters

partition

The partition coming from a clustering algorithm

Value

A list with six elements:

Rand

the Rand Index

ExpectedRand

expected value of Rand Index

AdjustedRand

Adjusted Rand Index

var_ari

variance of Rand Index

nari

nari

p-value

the p-value of the test

References

E_M. Qannari, p. Courcoux and Faye p. (2014) Significance test of the adjusted Rand index. Application to the free sorting task, Food Quality and Preference, (32)93-97

L. Hubert and p. Arabie (1985) Comparing partitions, Journal of Classification, 2, 193-218.

Author

Paola Tellaroli, <paola dot tellaroli at unipd dot it>; Philippe Courcoux, <philippe dot courcoux at oniris-nantes dot fr>

Examples

library(CrossClustering)

clusters <- iris[-5] |>
  dist() |>
  hclust(method = 'ward.D') |>
  cutree(k = 3)

ground_truth <- iris[[5]] |>
  as.numeric()

cc_test_ari(ground_truth, clusters)
#> $rand
#> [1] 0.8922595
#> 
#> $expected_rand
#> [1] 0.5525751
#> 
#> $adjusted_rand
#> [1] 0.7591987
#> 
#> $var_ari
#> [1] 8.607097e-05
#> 
#> $nari
#> [1] 81.83274
#> 
#> $p_value
#> [1] 0
#>