DOE - 2 Level or 3 Level?

preview_player
Показать описание
I get lots of questions on my e-mail from students who want to over collect data and go straight for a 3 level or more DOE. Here is my advice...
FREE DMAIC DOWNLOAD LINK...
Рекомендации по теме
Комментарии
Автор

I'm a fan here. Great work Paul. I work with chemicals and formulations to fabricate processed products that people buy at the supermarket. I've done it all: adhesives, paints, beverages, processed foods and cleaning products. So as you can guess mixture experiments are my thing. On the other hand, response surface methodology (rsm) is great for optimizing simple process variables: mixing speeds, loads, volumes, temperatures, pH, etc. Minitab and JMP are great, but I just love R. And soon found that applying separately mixture experiments for composition purposes and process experiments for productivity purposes, is just too much work. So the process-variable design is assigned to every mixture blend of a mixture design. In my world they call it the crossed mixture design and then it is coupled with RSM and multiple response optimization through a desirability function, which sounds about right. But is it right? Or are we overthinking this? What do you think?

victorst
Автор

Hi Paul, thank you for your videos I am learning a lot. I was wondering what would you advice if I have 4 categorical factors with 3 levels each -not ordinal (ie. apple types as the ingredient: gala, pink lady and honeycrisp). I understand the logic behind using only 2 levels in numerical factor cases however I am finding it difficult to understand what would be best for my case. Thank you in advance.

LVBS
Автор

Could you help in understanding the ANOVA model, how to execute it from starting, how to enter data and should we have to do replication?

kkk-lwox
Автор

Good afternoon Mr. Allen.

Do you know if DOE can be used to test the behavior of Artificial Intelligence system?

vinhk
Автор

A common mistake I see all the time, people try to shortcut and use the DOE test straight away to find the optimum parameters contained within the tested matrix instead of doing the all the necessary calculations, they simply pick the best results from the tested matrix. It could work but their results would be more efficient if they followed proper DOE methods.

digitalnightstudio