Consider an optimization problem with $n>3$ objectives. For handling this there exists often two approaches:

a) some weighting of the objectives,

b) fix an order of objectives and then optimize each one after another, allowing the objective of the previous optimization to only degrade by a small factor.

The weighting or the small factor is often difficult to estimate a priori. Assume now someone who wants to find a solution with a good balance between the objective, by manually tweaking these parameters and wants to identify cases by say 1% degrading in this objective I can gain 80% gain in a different objective.

What are useful plots/statistics/metrics for this, that help guide finding good parameters?



Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.