A light, and pure python package that aims to create a "port" between pyspark and python. It allows you write any feature transforms using the wrapper class located in the spark subpackage on standard spark DataFrame, save the entire feature transform pipeline in a file, and restore the pipeline in pure python program, saving a lot of time when you train a model in pyspark and want to use this model in a pure python program, for example, online production.

python setup.py

More to add