Any good suggestions for how to model human behavior? Preferably better than “monkey pressing the `your a nazi’ button.”

I have a model (a decision tree) modelling all possible behavior, and I want to easily enhance that to also 1) provide probabilities of picking certain paths, and 2) allow enhanced behavior.

I don’t want to force users to provide all probabilities (for example, I want to just be able to say that 80% of cases are fair-weather scenarios even if there’s 7 failure scenarios).

My models in principle all deal with finite domains, but I already have a bit of auxiliary information that will allow me to for example extend the domain of an enumeration { BELOW_25, AT_25, ABOVE_25 }.

I basically want to automatically generate realistic load tests from a functional test.