Functional (or sequence) data means that the input of the predictor is a function. It
can be represented by more or less noisy measurements, sampled
either on a regular grid or at random points. Examples include spectra or time
series. In an abstract sense,
functional data is caracterized by a 1D correlation structure.