Robust Clustering via NMF (non-negative matrix factorization)

When one rank is provided, NMF clustering will be performed. When multiple ranks are provided, this function will determine the optimal number of communities (K) by assessing the cophenetic coefficient across different values of K.

Usage

nmfClustering(
  mat,
  ranks = 10,
  nrun.per.rank = 30,
  min.coph = 0.95,
  nmf.method = "brunet",
  ncores = 1,
  plot = FALSE,
  seed = 2024,
  ...
)

Arguments

mat

Numeric matrix representing the input data for clustering analysis.

ranks

Numeric vector specifying the number of clusters to evaluate.

nrun.per.rank

Integer specifying the number of runs per rank for clustering.

min.coph

Numeric specifying the minimum cophenetic coefficient required for a rank to be optimal.

nmf.method

Character string specifying the method for NMF analysis.

ncores

Integer specifying the number of CPU cores to use for parallel processing.

plot

Logical indicating whether to plot the results.

seed

An integer used to seed the random number generator for NMF analysis.

...

Additional arguments to be passed to the nmf function.

Value

An NMFfitX1 object when only one rank is provided. A list containing the optimal number of communities (bestK), a list of NMFfitX1 objects (NMFfits), and a ggplot object (p) displaying the cophenetic coefficient across different values of K.

Examples

library(NMF)
library(SpatialEcoTyper)
mat <- matrix(rnorm(1000, 3), 20)
mat[mat<0] = 0

## Specify one rank
result <- nmfClustering(mat = mat, ranks = 3, nrun.per.rank = 3)
predict(result)
#>  [1] 3 2 2 3 2 2 2 2 3 1 1 1 1 3 1 2 3 3 3 2 2 2 2 1 2 1 3 3 3 3 2 3 1 3 3 2 3 2
#> [39] 1 2 2 3 3 1 3 2 1 3 3 3
#> attr(,"what")
#> [1] columns
#> Levels: 1 2 3

## Determine optimal ranks by testing multiple ranks
result <- nmfClustering(mat = mat, ranks = 2:5, nrun.per.rank = 3)
result$p

result$bestK
#> [1] 2
predict(result$NMFfits[[paste0("K.", result$bestK)]])
#>  [1] 1 1 2 1 1 1 1 1 2 2 2 2 1 2 2 1 1 2 2 1 1 1 2 2 1 2 1 2 1 1 1 2 2 2 1 1 2 1
#> [39] 2 1 2 1 1 2 2 1 2 2 2 2
#> attr(,"what")
#> [1] columns
#> Levels: 1 2