 |
|
|
|
The improper uniform returns P(x) = 1 for every value of x, all the time (and thus, log likelihood(x)=0). It has zero parameters. It is useful, for example, as an input to Bayesian updating, to represent a fully neutral prior.
The estimate routine is just a dummy that returns its input.
The draw function makes no sense, and therefore returns an error.
The uniform model. This is the two-parameter version of the uniform, expressing a uniform distribution over [a, b].
The MLE of this distribution is simply a = min(your data); b = max(your data). Primarily useful for the RNG, such as when you have a Uniform prior model.