ARMAsel for Identification of Univariate Measurement Data

For stationary random data, an automatic estimation algorithm selects a time series model whose spectrum is close to the Cramer-Rao lower bound. The parameters of that selected time series model accurately represent the spectral density and the autocovariance function of the data. That is all information for Gaussian data and also the most important information for arbitrarily distributed data. The improved computational speed gives the possibility to compute hundreds of candidate time series models, to select only one model and to forget the others. The three linear time series model types are: autoregressive (AR), moving average (MA) and the combined ARMA models. The ARMAsel algorithm computes models of the three types for a large number of candidate model orders. A single model type and order is selected from those candidates by looking for the smallest prediction error. That selected model includes precisely the statistically significant details that are present in the data, and no more. The automatic program can be incorporated in measurement instruments and also in protocols where the detection of changes is important