The FMM procedure makes the following assumptions in fitting statistical models:
-
The number of components k in the finite mixture is known a priori and is not a parameter to be estimated.
-
The parameters of the components are distinct a priori.
-
The observations are uncorrelated.
