irMF, inferential robust Matrix Factorization

[#1]

irMF standard version (version 8.1)

• Supports robust and regular singular value decomposition 
• Supports four algorithms for non-negative matrix factorization
• Supports sparsity options, convex and kernel variants (professional version only)
• Supports Non-Negative Tensor Factorization (professional version only)
• All irMF steps are recorded in irMF scripts. The addin can then be run in non-interactive mode and integrated in jsl, R or matlab pipelines (professional version only)
• Provides visualization, heatmaps, score plots of the factorization 
• Gives novel methods to judge the number of NMF factors 
• Effects a clustering of the rows and columns of the matrix 
• Provides association statistics between NMF clusters and actual groups, ROC curves 
• Gives novel methods of determining which columns can be used to predict class labels that is more powerful than traditional methods

The professional version of irMF is described in our website www.irmfpro.com

[ian_jmp]

It's in the documentation, but it's perhaps worth highlighting that this installs under the 'Analyze' menu. 'View > Add-Ins' provides a convenient link to the folder in which the PDF documentation lives.

[paulfogel]

Next version 8.0 will be posted soon. It will provide a new option to access the pdf doc from within the addin.