Global sensitivity analysis of the MILP model

Ongoing

Li Luo

• October 14, 2020 01:55

I would like to conduct global sensitivity analysis to test how the model objective and some other outputs change when some parameters change. The global sensitivity analysis requires large sampling size. For example, I will sample parameters such as transport cost and inventory cost 300 times using monto carlo simulation, which means I need to run the model 300 times. I tried the multiscenario method, but the result seems unsuccessful. Have anyone done the GSA on MILP model before? Thank you!

0

• Official comment

  

  Simranjit Kaur

  Gurobi Staff

  • December 06, 2024 00:41

  This post is more than three years old. Some information may not be up to date. For current information, please check the Gurobi Documentation or Knowledge Base. If you need more help, please create a new post in the community forum. Or why not try our AI Gurobot?.
• 

  Yuriy Zinchenko

  Gurobi Staff

  • October 15, 2020 22:32

  Attacking this problem with brute force sounds awfully expensive.  

  Perhaps you can try to explore some hybrid approach where for continuous variables you can extract standard sensitive info instead of probing those using the scenarios, e.g., optimize your MIP, fix the binaries and consider SAObjUp and SAObjLow sensitivity attributes for the resulting LP.  Of course, you would need to make your own judgment what this means for you from a modelling perspective.

  1
• 

  Federico Arroyo

  • November 28, 2021 19:56

  

  

  Hi Yuriy,

  How would you recommend to fix the binaries and continue further with the sensitivity analysis. Do I have to build a new model with the fixed binaries?

   

  Thank you for the help

  0

Post is closed for comments.