A nonlinear correlation measure for multivariable data set

The paper states that the mutual information carried by the rank sequences that are obtained from the original two sequences is a good measure of nonlinear correlation. Based on that, the nonlinear correlation information entropy (NCIE) is proposed for measuring the general relationship of a multivariable data set. NCIE uses a number in the closed interval [0, 1] to indicate the nonlinear correlation degree of the concerned multivariable data set, with 0 and 1 denotes the weakest and the strongest relationship, respectively. Data sets generated from an autoregressive model, a nonlinear chaotic Lorenz system, and a logistic function as known correlation sequences are analyzed using the proposed NCIE. The results testify the suitableness and correctness of the proposed concepts as nonlinear Correlation measure. (C) 2004 Elsevier B.V. All rights reserved.