irMF is one of a kind software for enhancing SAS JMP software for dealing with high dimensional, e.g. omic data. Core calculation modules being written in Python 3 can be used independently through a scikit-learn like interface (pip install nmtf), allowing full integration with production pipelines. The software can be customized to the needs of our customers.

NEW! Version 11 supports Hoyer's sparsity constrain with NMF and NTF implementations...

Show More

Modern methods such as microarrays, proteomics and metabolomics often produce datasets where there are many more predictor variables than observations. Research in these areas is often exploratory; even so, there is interest in statistical methods that accurately point to effects that are likely to replicate.  Using dimension reduction methods like the singular value decomposition and its close cousin, principal component analysis (PCA) produces score and loading matrices representing the rows and the columns of the original table and these matrices may be used for both prediction purposes and to gain structural understanding. When the data entries, e.g. gene expression levels, are necessarily nonnegative, the matrix factors meant to represent them should arguably also contain only nonnegative elements. This thinking, and the desire for parsimony, underlies the development of an attractive alternative, nonnegative matrix factorization, or NMF, which directly seeks matrix factors containing only nonnegative elements, rather than attempting to transform a loading or score matrix of mixed signs into one with only nonnegative elements. The resulting factorization often leads to substantial improvements in interpretability of the factors and in prediction of outcome.


“inferential robust Matrix Factorization”, or irMF,

  • Effects a clustering of the rows and columns of the matrix

  • Supports robust and regular singular value decomposition, with four algorithms for non-negative matrix factorization + sparsity options

  • Supports important variants of NMF, such as convex, kernel, PosNeg or semi-NMF

  • Allows for powerful, blind deconvolution of samples

  • Supports an extension to tensors, such as time-course data

  • Provides visualization, heatmaps, score plots of the factorization

  • Gives novel methods to judge the number of NMF factors

  • Provides fitted and residual matrices

  • 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 are more powerful than traditional methods


The standard version of irMF includes all NMF algorithms and utilities. However, irMF non-interactive mode is  part of the professional version only (see key features).

irMF, in its standard version, can be downloaded from SAS JMP Community


irMF is an exploratory software in the form of an Add-In to JMP (SAS Institute, Cary, NC, USA). Core calculation modules being written in Python 3 can be used independently through a scikit-learn like interface, allowing full integration with production pipelines. Running in the rich JMP analysis environment, irMF offers numerous data manipulations and visualizations.


irMF is an add-in to SAS JMP, customized to the needs of our customers. Interested parties should contact us to determine if the software, in its professional version, is of use and discuss a license. Support is most typically part of the license.

Example of licensing options