I like the idea. The tough part is there’s many input variables, and what are your clearly defined outputs? Yield? Thc %? This does lend itself to a “systems dynamics” problem, but at best you’ll only end up with bath tub style curves (graphs) that tell you more generalized answers. The REAL problem however is you need A LOT of input test data, as in growers, and I’m not aware of a forum that even has close to the kinds of numbers needed for this. How do I know? I spent many years working on systems dynamics problems using Latin hypercube and monte carlo sampling methods (advanced statistics).
TLDR; you need millions of growers to generate meaningful probabilities for these types of inputs & outputs.