MIclust_mpool

list of partitions, where one element of the list
corresponds to the clustering results for one imputed dataset. If more than
one clustering algorithm were used, each element if the list is a
dataframe, as obtained by <code>partition_generation()</code>.

list.part

Boolean, use <code>TRUE</code> to perform an additional consensus
from all partitions (ie. one consensus per clustering algorithm used, plus
one consensus of all partitions: mixing all clustering algorithms used).
This parameter is forced to <code>FALSE</code> if <code>length(algo)&lt;2</code>.

comb.cons

Numeric (in (0,1)) passed to internal function
<code>my_jack()</code>. Minimum <code>Jaccard index</code> value between partitions to
keep them for the consensus.

mcons.JAC.sel

Boolean, should <code>MultiCons()</code> tree be plotted?

plot.MIclust

Performs <code>MultiCons()</code> from a list of partitions

internal

Algorithms for (i) unsupervised learning for dataset with
missing data and/or left-censored data, using multiple imputation and
consensus clustering ; (ii) semi-supervised learning with a survival
endpoint (right-censored) for complete or incomplete datasets, using
multiple imputation and consensus clustering in the latter case. The
methods are described in Faucheux et al. (2021)
<doi:10.1002/bimj.201900366> and Faucheux et al. (2021)
<doi:10.1002/bimj.202000365>, respectively.

MIclust_mpool: `MultiCons` wrapper for imputed datasets

Description

Usage

Arguments

Value