hmm is a pure-Python module for constructing hidden Markov models. It provides the ability to create arbitrary HMMs of a specified topology, and to calculate the most probable path of states that explains a given sequence of observations using the Viterbi algorithm, or by enumerating every possible path (for small models and short observations). It also provides the ability to construct HMMs empirically, based on annotated observations.
hmm has been tested with Python version 2.6.
The source distribution for the most recent version can be obtained from the hmm project page by clicking on the Download ZIP button. The module can be installed with:
> python setup.py install
Since hmm is a work in progress, it's recommended to have the most recent version of the code.
hmm is a regular Python module; you import and invoke it from your own code. For a detailed usage guide and examples, please consult the user's guide.
Copyright (c) 2014 Michael Strosaker. See the LICENSE file for license rights and limitations (MIT).