InterJournal Complex Systems, 267
Status: Accepted
Manuscript Number: [267]
Submission Date: 990111
Revised On: 912
Entropy estimation from finite data samples
Author(s): Dirk Holste

Subject(s): CX.01, CX.05, CX.06, CX.07, CX.30, BG.00

Category: Brief Article


Entropy analysis is a common method to search for statistical patterns in measured symbol sequences or time series sampled from a underlying dynamical system. In order to obtain a comprehensive description, generalized Renyi (R) and Tsallis (T) entropies have been introduced. A practical problem lies in the numerical estimation of R and T from finite samples, which can lead to systematic and statistical errors. We focus on the problem of estimating R and T from limited data samples and derive the Bayesian estimators r and t, respectively. We compare r and t with the standard frequency-count estimators of R and T and find by numerical simulations that r and t reduce statistical estimation errors for processes such as generated by higher-order Markov models.

Retrieve Manuscript
Retrieve Previous Revision's Abstract
Submit referee report/comment

Public Comments: