A large number (\(n\)) of observations are assigned randomly into (\(xq\)) clusters. It is recommended to repeat Multimix runs with a number of different seeds to search for a log-likelihood maximum.
make_Z_random(D, seed = NULL)an object of class multimixSettings -- see
data_organise for more information.
a positive integer to use as a random number seed.
a matrix of dimension \(n\times q\) where
\(n\) is the number of observations in D$dframe
and \(q\) is the number of clusters in the model as specified
by D$numClusters.
Also consider making additional clusters from observations with low probabilities of belonging to any cluster in a previous clustering.
data(cancer.df)
D = data_organise(cancer.df, numClusters = 2)
Z = make_Z_random(D)
table(Z)
#> Z
#> 0 1
#> 475 475