Python Version of GLMnet

[vingkan]

The GLMnet model in R from the original project applies ridge (L2) regularization to the inspector-related features, but not any others. I searched for an equivalent version in Python:

• I could not find an option in Scikit Learn to pass an array of penalties for each feature.
• I could not find an option for L2 regularization in StatsModels.
• Finally, I used Glmnet in Python by Jerome Friedman, Trevor Hastie and Rob Tibshirani.

The Glmnet in Python package is only supported for 64-bit linux. I had to deal with a few issues to run the package on my Mac:

• According to this issue, the package on PyPI does match the master branch of the GitHub repository.
• I had to follow the steps in this issue to download another GLMnet.so file and move the Fortran files to the folder that the package expected them to be in.

After this, things ran smoothly. I referred to this Jupyter notebook for documentation.