Introduction to Two-Stage Stochastic Optimization (Conceptual)

[@shu-boyang4941]

The schematic picture for illustrating the idea of two-stage stochastic optimization is awesome. It makes everything so intuitive. Thank you for this great video.

[@TallysYunes]

Happy to hear it was helpful. Thank you for the feedback!

[@zinebelqabli3101]

​@@TallysYunes dear professor, thank you for your effort, you have explained very well stochastic programming, I want to know if you have other videos on : - stochastic dynamic programming - Non-linear programming - Multi-objective programming - Fuzzy programming - Quadratic programming - linear/integer/mixed linear programming - Robust optimization - Dynamic programming

[@giangdang3004]

This video actually helps me understand my assignment. Thank you for a very informative video.

[@TallysYunes]

Great to hear! You're welcome!

[@ahmedtawfiq3625]

Some of the best contents i have found from your channel 👏🏽👏🏽

[@TallysYunes]

Thank you!

[@imadeddineaziez7585]

Very helpful, thank you very much !

[@amirbaghban3068]

Thank you for your clear teaching way ...

[@sohomchat]

Thanks for the video. Can I use stochastic optimization to make a decision for a scenario that realized which I had not accounted for when I modeled the problem? In other words, is it only useful when I have complete knowledge on the finite number of possible scenarios?

[@TallysYunes]

Correct. The stage-1 solution needs to be aware, a priori, of the possible scenarios of stage 2 and their respective probabilities. If there's a potential "other" scenario with a non-zero chance of taking place, you could include that too at the beginning and estimate some data for it.

[@hullopes]

That was a really good class. Awesome! Tks!!

[@TallysYunes]

You're welcome! Glad you liked it!

[@zinebelqabli3101]

@@TallysYunes please sir, give me a concrete example of the application of this type of modeling approach in production planning, thank you

[@zinebelqabli3101]

dear professor, thank you for your effort, you have explained very well stochastic programming, I want to know if you have other videos on : - stochastic dynamic programming - Non-linear programming - Multi-objective programming - Fuzzy programming - Quadratic programming - linear/integer/mixed linear programming - Robust optimization - Dynamic programming

[@zinebelqabli3101]

​ @Tallys Yunes

[@TallysYunes]

I have only one video about stochastic DP. It's on the famous secretary problem: kzbin.info/www/bejne/Z6C5iGCMaJ2soNE. If you look through my Excel Models playlist, there are several videos on linear/integer programming. The other topics (non-linear and multiobjective) are on my to-do list for upcoming videos.

[@zinebelqabli3101]

@@TallysYunes thank you , and what is the difference between SDP( stochastic dynamic programming) and DP( dynamic programming) ?

[@TallysYunes]

Stochastic DP takes into account the probabilities of each of the possible outcomes (new states) after you take an action in the current state. In deterministic (or, non-stochastic) DP there are no probabilities involved. So, for example, you can use standard DP to compute the optimal solution to an integer knapsack problem.

[@zinebelqabli3101]

@@TallysYunes Are "stochastic programming" and "stochastic dynamic programming" the same or is there a difference between these two approaches?

[@qusayhamad7243]

thank you

[@prashantpant8501]

Hi professor, I have a small question. Say for example I have to prepare operation schedule for 2 machines- how much they should produce ( I have considered that there is no first stage variable the units are already commited=2). There is uncertainity of supply and demand, making in total S*D (scenarios). The problem is solved and we have one objective value, the solution for the production variable is different for S*D scenarios, but I need to provide one schedule for tomorrow. How to finalize one production schedule for tomorrow considering all the solutions from different scenarios?

[@TallysYunes]

What you describe doesn't sound right. The whole point of this type of stochastic optimization is that you must commit to making a decision *BEFORE* you know what will happen. You only know what can happen and the chances/likelihoods of each happening. Therefore, there must be stage-1 variables (today's decisions), or else the situation doesn't fit into a two-stage problem. In the production case, it must be that you need to commit to something today (maybe the schedule, i.e. how many units to make on each machine?). Perhaps there are additional details in your problem that you did not explain yet.

[@joelmorley5]

In this generalised example, where does the stochasticity appear?

[@TallysYunes]

I'm guessing you're referring to the generalized example in the picture at timestamp 8:13. The stochasticity is in the fact that I'm not sure what my tomorrow will look like (it could be one of three possible things) and I have to make a decision today before the actual tomorrow gets revealed to me.

[@tomdapchai]

thanks prof

[@tugbasarac1087]

👍