In an attempt to automate several steps of a process, an AI based tool is being considered. I’ve been tasked to verify the robustness of same.  The output would be expressed in accuracy.  The critical factors that would be at play while using the AI tool have been narrowed down to 3 factors having 2 levels each.  With 2 replications, I am currently in the process of conducting a full-factorial experiment with 16 runs.  With no time and cost constraints, how many replications should I go for so that the factors are best optimized and I have more confidence in the output that is generated?

Hello Pradeep

 

2 replicates are fine. It will take care of noise factors during the experimentations.

 

Regards

Mayank Gupta

Thank you, Mayank!!!