Scenario: I have 8 compounds A-H, for each compound I have 5 runs* of time-temperature data with a known category. I need to be able to compare a single data run of an unknown category to those existing runs and assign a category to that run.
A few things going on which may also need accounted for:
1) My start times (when heating began vs the data run started) do not all align.
2) Not all runs end at the same temperature (for safety reasons)
3) Some of these compounds are very similar and may not be distinguishable (whether I can or not is the purpose of this research)
I've attached my data as well as the image of the plot overlay.
I'm looking for help on what tool/function to use to go about setting this up. Right now, I can run things like analyze->multivariate to get a table with each data run against all the others, but this does not allow for comparing to the set of data. I also tried analyze - > cluster variables, but it only gives me two clusters and I don't know how to tell it to give me more. Nor can I see a way to manually train the initial clusters. Other options I tried appeared to be comparing individual row values, not the curve as a whole, which doesn't take into account the larger picture, which is what I need to be comparing.
I'm happy to dig deeper into the documentation on how to use the suggested tool/method/function, but I need advice on a starting place. I have access through my university so I should have access to most if not all options out there. I also have access to other analytics packages (including SAS) if there is a better option from one of those.
Scripts are ok, but I'd prefer not if possible since I'm new to JMP and under a time constraint.
*My focus right now is on getting a process in place that can at least separate out the obvious non-matches before I spend time and resources on additional data runs.
