From what you’re saying is that you want something Bayesian, but you can’t define the likelihood. For such cases there is approximate Bayesian computation (see abc), where in place of likelihood you use some summary statistics.
Tim
- 141.2k
- 26
- 270
- 512