py2ds2

Python package to create a SAS DS2 scoring package from a pickled scikit-learn pipeline.

Installation

1. Clone this repo
2. cd py2ds2
3. pip install py2ds2

Usage

import py2ds2
py2ds2.create_from_pickle(pickle_filename, X, model_function='CLASSIFICATION')

Creates a SAS DS2 scoring package from a Python pickle file.

The pickle file should be a sci-kit learn pipeline that was fit on the pandas DataFrame X. A python scoring file, a SAS DS2 scoring file that wraps the python scoring file, and input/output variable info files are all created from the pickled scikit-learn pipeline.

See the examples directory to see an example of a jupyter notebook that builds a sci-kit learn ML pipeline, pickles it, then uses py2ds2 to create the necessary SAS scoring files in pymodels/hmeq2