An observation is a single frame of data in a sequential pattern. En entire pattern sample is therefore a sequence of observations.

In the AME Patterns library observations are held in types that model the Observation concept. For example, when dealing with patterns of numbers, observations might be held in a double.

Pattern modeling often relies on observation probability distributions. An observation probability distributions is simply a probability function (probability density function) over the observation space. In the AME Patterns library, distributions are held in types that model the ObservationDistribution concept.