At a certain point you kind of just have to do the work of roasting, tasting, reviewing the data, and iterating on an idea. Progressive roasting exercises are an incredibly powerful training tool for that, but it can be difficult to justify at home roasting scale (much of the equipment is completely impractical for it and it's something that works out to be cheaper the more of the same coffee you're getting so financially this is unappealing at less than half bag quantities).

Another challenge when it comes to considerations of rate shaping is that temperature measurements are not the same across different machines. You can almost always find equivalent temperatures to construct a function taking temperatures on one machine and getting the temperatures that you'd read on a different one, but unless that looks like y = x + c (constant offset) it's going to be possible to construct a roasting plan that looks like a declining rate when viewed through the lens of one machine and not on a different one. It's not uncommon to see larger differences among machines at hotter temperatures (people who do this work more than I do have reported that sometimes the constant offset case exists, but in my experience it's almost always some kind of non-linear curve), which makes rate based considerations considerably less useful (or at least obviously not universal). Time within ranges combined with a measured roast level are, I think, a more useful framing unless everybody is talking about roasts on the same machine, though it's hard to justify the expense of that for a home roaster (or even most small commercial roasters).