hsmmlearn
is a library for unsupervised learning of hidden semi-Markov
models with explicit durations. It is a port of the
hsmm package for R, and in
fact wraps the same underlying C++ library.
hsmmlearn
borrows its name and the design of its api from
hmmlearn.
hsmmlearn
supports Python 3.6 and up. After cloning the
repository, you can install the package by running
pip install .
Note the dot (.
) at the end of the command, which is part of the command. You
will need a C++ compiler to build and install the package.
To run the unit tests, do
python -m unittest discover -v hsmmlearn
The documentation for hsmmlearn
is a work in progress. To build the docs,
first install the doc requirements, then run Sphinx:
cd docs
pip install -r doc_requirements.txt
make html
If everything goes well, the documentation should be in docs/_build/html
.
Some of the documentation comes as jupyter notebooks, which can be found in the
notebooks/
folder. Sphinx ingests these, and produces rst documents out of
them. If you end up modifying the notebooks, run make notebooks
in the
documentation folder and check in the output.
hsmmlearn incorporates a significant amount of code from R's hsmm package, and is therefore released under the GPL, version 3.0.