Using Entropy in Time Series Analysis

Abstract

Results of experiments in the field of biophysics are often presented as time series obtained with low resolution and not always of great length. In particular, in studies of the effects of various physico-chemical factors on bilayer lipid membranes, transmembrane ion currents and their fluctuations are usually measured. In this case, the mean values and variances of the currents may not differ significantly, making it difficult to determine the nature and degree of impact based on them. Therefore, the development of approaches to time series analysis has never ceased. Attempts to use the entropy of random variable distributions in such analysis have been made for a long time, but in practical work, these approaches have been difficult to implement, especially due to the requirements for the length of the series and the absence of noise. In recent decades, there have been significant changes in this area, and many new methods of time series analysis using various modifications of entropy have been proposed. In this regard, there is a need for a summary of methods based on entropy calculation, indicating their advantages and disadvantages. This is the goal of the proposed brief review of entropy-based methods for analyzing scalar time series, which can be useful in analyzing experimental data. The review considers only some of the basic approaches on which further algorithmic improvements are based. The concept of entropy sometimes causes difficulties for students, so the review can also be useful for educational purposes.