Welcome to the toolbox for tensor decompositions, statistical analysis, visualisation, feature extraction, regression and non-linear classification of multi-dimensional data. Not sure you need this toolbox? Give it a try on mybinder.org without installation.
Table of Contents
There are two options available:
Install
hottbox
as it is from pypi.org by executing:# Create virtual environment based on python 3.7 pipenv --python 3.7 # Install hottbox from pypi pipenv install hottbox
Alternatively, you can clone the source code which you can find on our GitHub repository and install
hottbox
in editable mode:# Clone source code and cd into it git clone https://github.com/hottbox/hottbox.git cd hottbox # Create virtual environment based on python 3.7 pipenv --python 3.7 # Install hottbox from source pipenv install -e .
This will allow you to modify the source code in the way it will suit your needs. Additionally, you will be on top of the latest changes and will be able to start using new stable features which are located on develop branch until the official release. The list of provisional changes for the next release (and the CI status) can be also be found on develop branch in CHANGELOG file.
NOTE: To manage working environment, we use pipenv. This tools is thought of as a successor of
pip
and essentially usespip
andvirtualenv
under the hood. If you haven't made a switch yet (or don't want to) then:- Instead of
pipenv install
simply usepip install
- Instead of
pipenv run ...
make sure that you are in correct virtual environment
- Instead of
hottbox
is under active development, therefore, if you have chosen the second installation
option, it is advisable to run tests in order to make sure that your
current version of hottbox
is stable. First, you will need to install pytest
and pytest-cov
packages:
pipenv install -e '[.tests]'
To run tests, simply execute inside the main directory:
# Runs pytest within a virtual environment
pipenv run pytest -v --cov hottbox
Please check out our repository with tutorials on hottbox
api
and theoretical background on multi-linear algebra and tensor decompositions.
We welcome new contributors of all experience levels. Detailed guidelines can be found on our web site.