Traditional data analysis in a batch environment often does and/or cannot take into account all data of the process parameters recorded over time. You then take the startin value, mean or end value and hope to find relationships between these values and your response. There is much space to miss information, e.g. variation over time of these recorded parameters between batches.
At the end you are wondering why the process varies if all parameters stay in specs and the mean or starting value is almost the same value for every batch.
Take a look at why the Functional Data Explorer in JMP Pro most likely will help you in those situations. No matter if you want to have a better classification of your batches or understand the relationship between variation over time in your process parameter with the response like yield.
You want to know more, get interested or have situations you think this is valuable. I'm happy to discuss this with you. You have situations like that and used FDE already, please let me know and comment.