A statistical model is identifiable when distinct parameter values determine distinct probability distributions. In a nonidentifiable parameterization, data can at most estimate combinations of parameters that remain unchanged along each equivalent parameter set.
New to topics? Read the docs here!