A literal implementation of the LARS algorithm described by Efron, Hastie,
Johnstone, and Tibshirani (2004). This implementation is much less complex than
the one in scikits.learn, which I think might be useful for pedagogical
purposes. For doing "real" regression problems, though, I'd recommend using the
scikits.learn implementation.
Install with pip
pip install lmj.lars
Or directly from the source
git clone http://github.com/lmjohns3/py-lars
cd py-lars
python setup.py install