The previously introduced (weighted) least squares principle is deterministic, since the underlying criterion does not make use of the probability distribution of the observables. Of course, the resulting (weighted) least squares estimator is stochastic, since it is a function of the stochastic observables as explained in the previous unit.

We will now introduce a different estimation principle, which is probabilistic. This means that the stochastic properties are used to derive the estimator, which is called the Best Linear Unbiased Estimator.