This experiment illustrates the use of Execute Python script to train and score Scikit-Learn models
Execute Python Script currently does not allow you to return anything except a Pandas data frame. To return trained models that can be used in a subsequent scoring experiment, you must first serialize it to a string via the `pickle` module and pack that into a data frame. The scoring experiment can then deserialize or unpickle the model and score it on the test dataset. Pickled models can occasionally get fairly large. You can compress them via the `zlib` module and return the compressed output back to AzureML.