The German Tank Problem

Something interesting happens when we make this adjustment to the estimator — the average estimate across all simulations exactly matches the total number of tanks, 1000. Our first estimator, the maximum serial number of the captured tanks, turns out to be the maximum likelihood estimate (MLE), and its adjustment is the so-called minimum variance unbiased estimator (MVUE). Instead of setting a fixed value for the total number of tanks, with probabilistic programming, we define a prior distribution that encodes our apriori beliefs and knowledge about the parameter(s) of interest.

Source: www.eadan.net