FIxes and modifications for the extraction and prediction jobs
The spark.sql query has been replaced with a more pythonic implementation gotten from revertrisk-datasets
Exception handling has been added to the predict UDF for error thrown on null or empty fields in the dataframe