1. XGBoost library
2. scikit-learn

1. Download XGBoost and sklearn libraries to all nodes of your Greenplum cluster (using gpscp -f hostfile <source> =:<destination>)
2. Compile and install XGBoost and sklearn libraries on all nodes (using gpssh -f hostfile followed by the compile and install commands)
3. Run the above SQL file (it will create a schema called xgbdemo).
4. Invoke the UDFs as shown in the sample snippet.

Since the XGBoost implementation in https://github.com/dmlc/xgboost is not Python 2.6 compatible, I recommend you clone my version from https://github.com/vatsan/xgboost and use it instead (Python 2.6 compatible, will work with PL/Python on Greenplum/HAWQ).