I’m trying to code up a model for log(pressure) vs. some function of temperature for a set of chemicals, and I know that the model before and after my triple point (which is continuous) will be different. For example one model combination is:
ln(p) = a1 + a2/ T + a3×ln( T ) + a4× T + a5× T^2 + a6/ (T^ 2) + a7 × T^6, T < gamma
ln(p) = a + b/ T + c×ln(T) + d× T^2 + e/ (T ^ 2), T >= gamma
My natural instinct is to use a change point model, but is there another way to do this? Ideally a method that doesn’t result in a discontinuous model.