A complexity score derived from principal components analysis of nonlinearorder measures

The generation of a global "complexity" score for numerical series was deri ved from a principal components analysis of a group of nonlinear measures o f experimental as well simulated series. The concept of complexity was demo nstrated to be independent from other descriptors of ordered series such as the amount of variance, the departure from normality and the relative nons tationarity; and to be mainly related to the number of independent elements (or operations) needed to synthesize the series. The possibility of having a univocal ranking of complexity for diverse series opens the way to a wid er application of dynamical systems concepts in empirical sciences. (C) 200 1 Elsevier Science B.V. All rights reserved.