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; } } }

Prediction profiler - how to find the most stable operating point rather than maximise the response

cbaril

Hello,

Using the prediction profiler one can determine the conditions maximizing the Response.

Is it possible to find as well the most stable operating points?

As there are many interaction effects, it is too difficult to identify visually.

Many thanks in advance

Best regards,

Claire

Accepted answers

David_Burnham

You might find this blog post of interest: Finding the best process operating conditions using 2 optimisation approaches

All comments

P_Bartell

Sure. Depending on the type of data and responses you have there are several options. The trick will be to find a place where the slopes of the various factors are relatively flat when certain levels are chosen for the factors that are highly interactive. Depending on how many interactive effects you have, you might want to, within the Prediction Profiler, lock those factors at levels where the slopes are flattest. Then use the co-optimizer and desirability function to find your most robust position. Another thought is to use a Monte Carlo simulator with the interactive factors mean and variation set near their flattest points. You may have to give up some bias to target for a less variable response. Lots to consider.

statman

Claire, Some questions for you:

1. I don't see the data set attached, so what is the data set? Is it from an experiment?

2. How many levels are there for the factors in the model?

3. Do you have a way of quantifying "stability"? Perhaps a measure of variability over some time period.

4. If you don't have one, can you create one from your data set?

In essence, the response variable you are using is likely providing a model for the mean of the response. What you may be interested in is the variation in the response. If you can create this, then you can use the profiler and interaction plots to determine which factors affect that response variable.

What you are looking for is a plateau on the theoretical "response surface". This is a good idea as this is one aspect of robustness.

David_Burnham

You might find this blog post of interest: Finding the best process operating conditions using 2 optimisation approaches

cbaril

This is a great tutorial! It's exactly what I was looking for! Many thanks

cbaril

Thank you for your quick reply!

The data set is confidential, the factors are continuous and the DoE had 3 Levels. For us stability is when the response varies little when varying the factor settings within the characterized range.

The tutorial in the accepted solution helped me to find the most stable operating point and to complete the analysis.

cbaril

Hello,

Many thanks for the quick reply!

I used the Monte Carlo simulator, it was exactly what I was looking for! Thanks!

David_Burnham

Great to hear. I forgot to acknowledge @robert_anderson as the author of the tutorial.