Originally posted by: kc78

Hi

I would like to find out is there any way to find out which variables in a MILP can reduce the variability of the system the most besides what i can think of in a brute-force approach, i.e. fixing each of the variable one by one at different intervals and calculating the sum of the min/max of the other variables. Or are there methods to sample the solution space of MILPs to do some statistical analysis of variance of the system?

Thank you.
#CPLEXOptimizers
#DecisionOptimization

Originally posted by: SystemAdmin

Do you want to solve such a "variability" problem exactly? Or would the solution pool, maybe with a diversity filter, would provide you some good estimate?
#CPLEXOptimizers
#DecisionOptimization

Originally posted by: kc78

Hi Tobias

I would like to find out which variables if i can reduce their variability, i.e. their bounds, would significantly reduce the "overall variability" of the system as defined as the max-min values of a set of variables in the LP or MILP.

What type of sensitivity analysis methods can i employ?

Thank you.
#CPLEXOptimizers
#DecisionOptimization

Originally posted by: SystemAdmin

Sorry, I do not know any efficient way to calculate such a measure.
Maybe you should just sample the solution space (using the populate method) and then get some estimate of the variability induced by the variables directly from this sample.

Tobias
#CPLEXOptimizers
#DecisionOptimization