benchmarks-functions: Benchmarks functions for clustering

In MoMPCA: Inference and Clustering for Mixture of Multinomial Principal Component Analysis

Description

These are wrapper to other methods for the clustering of count data. They can be used to initialize the clustering. It is also possible to implement your own benchmark function depending on other packages.

Usage

benchmark.random(dtm, Q, ...)

benchmark.kmeans_lda(dtm, Q, K, nruns = 1, ...)

Arguments

dtm	an S4 object of class mmpcaClust
Q	The number of clusters
...	Some argument to be consistent with the function's skeleton : K and nruns are optional arguments for some of them.
K	Number of topics (dimension of the latent space).
nruns	Number of restart of the kmeans() algorithm.

Value

A vector of size equal to the number of row of dtm, containing a Q-clustering

benchmark.random

Random initialisation of the clustering. Arguments K and nruns are unused

benchmarks.kmeans_lda

Cluster the matrix theta obtained by a topicmodels LDA with K topics

MoMPCA documentation built on Jan. 21, 2021, 5:09 p.m.