This is useful to compare the actual results to the ideal results we would obtain if the model could predict the exact misdirection of each event. We could see the maximum room for improvement and test the performance of different partitions while the partition method itself is still not completely optimized.
This is useful to compare the actual results to the ideal results we would obtain if the model could predict the exact misdirection of each event. We could see the maximum room for improvement and test the performance of different partitions while the partition method itself is still not completely optimized.