time-series-nmf 0.1.0.dev0

time-series-nmf

time-series-nmf is a Python package implementing non-negative matrix factorization for time series data. Currently, it supports a version with Tikhonov regularization and sparse constraints as proposed by Fabregat R.. et. al. and implemented in Matlab in https://github.com/raimon-fa/palm-nmf. It uses a PALM optimization scheme. We plan to add other models and optimization algorithms.

This work is under continuous development.

Installation:

# from github (latest)
pip install git+https://github.com/valentina-s/time-series-nmf

# from pypi (stable)
pip install time-series-nmf 

Getting started:

import tsnmf

# generate some data
from numpy.random import rand
data = rand(100,1000)

# fit time series nmf to data
model = tsnmf.smoothNMF(n_components=5)
model.fit(data)

# outputs
model.W
model.H

Acknowledgements

This work has been supported by NSF award #1849930 and the Gordon and Betty Moore and Alfred P. Sloan Foundations Data Science Environments grant (MSDSE).