Test Scalability of Spark BatteryAnalysis

[Mwegert]

Expected Behavior

The Spark program should work even if portions of the dataset are too large to fit in memory.

Actual Behavior

We suspect that there may be scalability issues in ModelFunction. We have provided ModelFunctionDataset, which resolves this but ran slower when we tested it. If there are issues, simply replace ModelFunction with ModelFunctionDataset in the Driver program. Look into optimization of the dataset logic in ModelFunctionDataset.

Steps to Reproduce the Problem

1. Generate gigabytes of data
2. Test the code as-is on an EMR cluster.
3. Determine if an error occurs.