How to present comparison in sensitive distance performance?

[alecgunny]

Should it be:

• Plot curves over one another
• Ratios
• Percent difference
• Difference in terms of number of number of standard deviations of control curve

Definitely applies to:

• reseeded training
• inference sampling rate

Not sure if applies to:

• hyperparameter settings
• Interval vs. original test set Do we just plot the SV curves against one another for these?