Fit a simple multinomial model for count data, in
which each sample (i.e., a row of the data matrix X)
is assigned to a cluster. Under this simple multinomial model,
\(x_{ij}\) assigned to cluster \(k\) is multinomial with sample
size \(s_i = x_{i1} + ... + x_{im}\) and multinomial
probabilities \(p_{1k}, ..., p_{mk}\). This is a special case of
the multinomial topic model in which all the mixture proportions
are either 0 or 1. The maximum-likelihood estimates (MLEs) of the
multinomial probabilities have a closed-form solution; no
iterative algorithm is needed to fit this simple model.
fit_multinom_model(cluster, X, verbose = c("none", "detailed"), ...)A factor specifying a grouping, or clustering, of
the rows of X; e.g., the “cluster” output from
kmeans.
The n x m matrix of counts; all entries of X should be
non-negative. It can be a sparse matrix (class "dgCMatrix")
or dense matrix (class "matrix"), with some exceptions (see
‘Details’).
This is passed as the “verbose” argument in
the call to init_poisson_nmf.
Additional arguments passed to
init_poisson_nmf.
A multinomial topic model fit.