Skip to content

Latest commit

 

History

History
35 lines (27 loc) · 587 Bytes

README.md

File metadata and controls

35 lines (27 loc) · 587 Bytes

nmf_py

(Gaussian)NMF(Non-Negative Matrix Factorization) implementation on python. This code is designed to handle large sparse matrix.

usage.

nmf = NMF()
// random sparse matrix
A = nmf.rand_sparse_matrix(100000, 1000, 0.01)
nmf.setup(A, k=10)
nmf.run(iter_num=100)
// hard clustering
nmf.clusters()

performance

spec

  • macbook pro retina 15inch
  • mac 10.8.2
  • 2.7GHz corei7
  • 16GB memory
  • scipy superpack

setup

  • sparse matrix size: 1000000 x 10000
  • density: 0.005
  • hidden variable num: 10
  • iter_num: 100

-> 147.484 sec = 2.3 min