This version has additional features such as :

1. Distributed data pruning as preprocessing for zero row/column removal followed by unpruning.
2. Checkpoint to track state of NMFk so that the session starts from that state after restart.

pyDNMFk/ Distributed pyNMFk is a software package for applying non-negative matrix factorization in a distributed memory to large datasets. It has the ability to minimize the difference between reconstructed data and the original data through various norms (Frobenious, KL-divergence). The current implementation utilizes optimization tools such as multiplicative updates, HALS, BCD and BPP. Additionally, the Custom Clustering algorithm allows for automated determination for the number of Latent features.