UML: PERFORMANCE - NMAI-lab/JLOAF GitHub Wiki

Performance is responsible for testing how well an agent has learned expert behavior and for providing statistics on the agent's performance.

PerformanceEvaluator is an abstract class that is responsible for determining the performance of an agent. It has two abstract methods that need to be implemented in any subclass: CreateAgent() and createArrayOfCaseBaseNames(). Take a look at the docstring to see how to implement them.

PerformanceEvaluatorMethod() performs leave one out testing on a list of casebases. It records the stats for each cycle in a statistics object, adds the statistics object to a list, and at the end calculates the mean and standard deviation. It also writes the results to the console, a .csv file and a SQLlite database.