About the time series analysis based on the forbidden permutation patterns searching methods
Maksim L. Korneevets,1, Andrey V. Starodubov,1,2 1 Saratov State University, Saratov, Russia 2 Saratov Branch, Institute of Radio Engineering and Electronics, RAS, Saratov, Russia
Abstract
The development of methods that can distinguish between noise and chaotic signals is currently an important task for scientists and engineers from various fields of knowledge. Analysis of the nature and structure of time series allows predicting heart or epileptic seizures, the breakdown of various engines, the behavior of financial markets and many more adverse events before they occur. Earlier, an original approach to the analysis of time series of systems of a very different nature was proposed, which is based on the search for permutation patterns and the calculation of the entropy of permutations. This method has already been successfully used to analyze signals of various nature. The purpose of this work is to develop an algorithm for analyzing time series using methods for finding and counting forbidden permutation patterns, which is a small modification of the classical method. As a result of the carried out studies, it was found that the developed approach makes it possible to identify the moments at which anomalous events begin to occur in a studied system.
Speaker
Maksim Korneevets
Saratov State University
Russia
Discussion
Ask question