This is not a specific question regarding how to use FDE, or how to interpret any of its output, more so a question around implementing it in a practical way and what that workflow looks like.
JMPs documentation on the FDE platform suggests that the 2 main ways to use FDE is for exploratory analysis (so in a manufacturing context let's call that process monitoring through the FPC control charts) and predictive modeling. I am looking for any guidance/advice on how to build your model, then more importantly, applying it to future batches based on their functional form. Do you have to just add the new batches or batches to your original data table and re-run the model fitting to get the FPCs for the new batches? Is there a way to project a new batch onto the space of your originally built model and get a prediction for some response of interest? What is the most efficient way to go about this?
Also, does anyone have any advice on how to monitor the functional form of a continuous stream of batches coming in from some manufacturing process? Do you continuously run the model fitting and "re-check" the FPC control charts each time?
I apologize for the rambling nature of the questions, so I will try to sum them up with 1. While the FDE platform produces interesting results and is very helpful for one-off exploratory analyses or DoE output, I am just looking for some recommendations on how to practically implement its modeling/monitoring capabilities in a typical business workflow. Thanks
