Perform multiple hierachical clustering on random subsets of a dataset
iterative.hclust(x, seeds = 1:100, row.rate = 0.3, col.rate = 0.1,
max.cluster = 10L, ret.height = FALSE, hc.method = function(x, PCs
= 1:6, ...) { hclust(dist(prcomp(x, rank. = max(PCs))$x[, PCs, drop =
FALSE]), ...) }, ...)
the numeric matrix containing the data to cluster (one instance per row)
a vector of random seed to use.
numeric value in [0,1] to specify the proportion of instance (resp. feature) to subset at each random iteration.
upper bound on the number of expected cluster (can by +Inf).
a logical to specify whether the average merging height should be returned.
a clustering method of arity 1, taking as input a random subset of the input matrix x and returning an hclust object
additional arguments are passed to the hc.method
a list of 3 square matrices N,H,K of size nrow(x): N is the number of time each pair of instance as been seen in the random subsets; H is the corresponding sum of heights for the pairs; K is the sum of the number of split possible that still preserve the two samples into the same cluster.