We have developed a Benders decomposition algorithm for an inventory-routing problem. We have considered that optimality and feasibility gaps could get different values. Therefore, choosing an appropriate combination of these two values over instances of various sizes is crucial.
To determine these values, I have defined six levels for each gap. Then, I have generated ten instances of different sizes and solved them with all the possible combinations of these gaps for once to record the CPU time. For instance, the following image shows the results obtained for one of the instances over different levels of two gaps:
https://ibb.co/wKjzT4V
Taking what mentioned earlier into account, is there any statistical test to rank these combinations over all instances to choose the best one (or at least recognize the best ones)?
If you need more information or clarification, please let me know.
I want to thank you in advance for the privilege of your time and your help.
Stumble It!
Tweet This
Bookmark on Delicious
Share on Facebook
Share on MySpace