Maximum information preservable by supervised hierarchical lumping

Supervised analogue of maximum_mutual_information_hierarchical(), exploiting hierarchical structure to divide the problem into independent sub-problems.

Usage

maximum_mutual_information_hierarchical_supervised(
  joint_counts,
  threshold,
  clusters,
  verbose = FALSE
)

Arguments

joint_counts

Named numeric matrix as described in maximum_mutual_information_nominal_supervised().

threshold

Minimum number of samples each lumped level must contain.

clusters

List of character vectors; each vector names the levels that may be grouped together.

verbose

Whether to print diagnostic messages. Default: FALSE.

Value

A list containing information about the optimal lumping:

mutual_information

Mutual information between the lumped covariate and the outcome, in nats.

loss

Mutual information lost by the lumping.

lumping

A list of character vectors, where each vector contains the names of the original levels that have been lumped together.

See also

maximum_mutual_information_hierarchical() for the unsupervised version.

maximum_mutual_information_nominal_supervised() for the general version.

Author

Daan Koning

Examples

joint_counts <- matrix(
  c(1, 1, 2, 0, 1, 2, 1, 1),
  nrow = 4,
  dimnames = list(c("Utrecht", "Friesland", "Bayern", "Sachsen"), c("y0", "y1"))
)
clusters <- list(c("Utrecht", "Friesland"), c("Bayern", "Sachsen"))
maximum_mutual_information_hierarchical_supervised(joint_counts, threshold = 3, clusters = clusters)
#> $mutual_information
#> [1] 0.005000792
#> 
#> $loss
#> [1] 0.1035853
#> 
#> $lumping
#> $lumping[[1]]
#> [1] "Utrecht"   "Friesland"
#> 
#> $lumping[[2]]
#> [1] "Bayern"  "Sachsen"
#> 
#>