nav[aria-label="Primary Navigation"] { padding: 0; & ul { list-style: none; width: 100%; display: flex; flex-direction: row; justify-content: start; align-items: start; gap: 30px; padding: 0; & li { margin: 0; } & ul li { list-style: none; } } }

Could JMP in 1 step optimize the uncertainty of multiple discrete inputs or integer values ?

Marco1

Hello community!!!

A query, related to multi-objective optimization... Will JMP be able to optimize the uncertainty of the random discrete inputs (integers) in "1 single step"?... that is, JMP will be able to find the best combination of inputs to use ( whole numbers) and the risk associated with each strategy?...and thus be able to search for strategies that allow minimizing risks while achieving the objectives?

That is, take any optimization problem and have JMP replace the uncertain values with probability distribution functions (integers - discrete) representing a range of possible values and for each trial solution found by JMP during optimization, JMP run a Monte Carlo simulation, finding the combination of adjustable cells that provides the best simulation results that minimize risk in the face of uncertainty?.

Accepted answers

All comments

Byron_JMP

Yes,

https://www.youtube.com/watch?v=4ZoPGjhu6SQ

Byron_JMP

and a longer response to yes

https://www.youtube.com/watch?v=9KZ7Ns3CQzU

Marco1

Hi Byron,

Thank you for your answer, one question:

- Is it the same to add noise factors....or to use the simulation experiment?...what is the difference?

For example, if you need to predict the stock or amount of inputs to buy for a restaurant (optimize uncertainty).

Cheers,

Marco

P_Bartell

A few of your statements I think need some clarification. While my former (I'm retired) @Byron_JMP 's comments are on the spot...the one stop shop for all your optimization needs in JMP resides in the Prediction Profiler, a few of your statements I think need some clarification or additional commentary.

1. "Will JMP be able to optimize the uncertainty of the random discrete inputs..." The answer to this is 'no.' if, by 'uncertainty' you mean the distributional properties of the input variables. JMP does not 'optimize uncertainty of the random discrete inputs'. You specify the uncertainty of the random inputs. JMP finds an optimal solution for a number of user specified inputs...among them...distribution shape of the inputs, central tendency, and limits of exploration across the prediction space.

2. "...search for strategies that allow minimizing risks while achieving the objectives." What are the risks? Practical, statistical, compliance, etc. Evaluating a minimal risk solution is impossible without knowing the exact nature of the risks and more importantly consequences associated with failure/noncompliance, etc.

3. What do you mean by '...multi-objective optimization.'? In a modeling sense do you mean multiple 'y' variables? Or do you mean something else, like 'on aim with minimum variance for a single 'y'. Or for that matter, 'on aim with minimum variance for multiple 'y's.'? Tactics you take within the Prediction Profiler will vary based on these answers. What role might JMP desirability function capabilities play here?

Monte Carlo simulation is a rich set of capabilities within JMP as well. Exact tactics you take will depend on much of what I've written above. I encourage you to finish watching all the links provided by @Byron_JMP

Marco1

Hi P_Bartell,

Thanks for your answer, there is a very interesting software called RISKOptimizer - Optimization with Simulation ( https://www.palisade.com/riskoptimizer/?gclid=CjwKCAiAhKycBhAQEiwAgf19egKupUCflDVglQVv3We-onUI-GrXcWJs_YATP83K1T__dE6QCQweaRoCn7IQAvD ) and I would like to know the difference J_BWE .
Cheers!

Marco

P_Bartell

I suggest reading the JMP documentation end to end regarding the Prediction Profiler and compare the two applications yourself.

Marco1

Hi P_Bartell,

Thanks for your answer.

Cheers,

Marco