A Method for the computation of entropy in the Recurrence Quantification Analysis of categorical time series
A Method for the computation of entropy in the Recurrence Quantification Analysis of categorical time series
复制标题
DOI：
10.1016/j.physa.2018.08.058
复制
发表时间：
2018-12-15
影响因子：
3.3
通讯作者：
Leonardi, Giuseppe
中科院分区：
文献类型：
--
作者：
Leonardi, Giuseppe
在这项工作中，我提出了一种新的方法，用于计算信息熵的递归图时，分析的时间序列是分类的性质。在这种情况下，通常在选择分析参数时会有一个简化，在这个意义上，通常不假设多维空间中的嵌入，并且递归仅限于数字编码类别的精确匹配（等价）。然而，这样一个简化的参数化带来了一些显着的变化所获得的递归图，这对标准动力学措施的提取的后果的外观。具体而言，分类递归图通常由矩形结构而不是线结构（对角线和水平/垂直）组成，递归量化度量最初是在其上提出的。从这一观察开始，我考虑替代计算程序来提取分类情况下的无偏熵度量，用模拟数据显示这种选择的可行性（C）2018 Elsevier B. V.保留所有权利。
In this work, I propose a new method for the computation of informational entropy from Recurrence Plots when the analyzed time series are categorical in nature. In such cases, there is typically a simplification in choosing the parameters of the analysis, in the sense that no embedding in multidimensional space is usually assumed and that recurrence is restricted to exact matching (equivalence) of the numerically coded categories. However, such a simplified parameterization brings about some notable changes in the appearance of the obtained Recurrence Plots, which has consequences for the extraction of the standard dynamical measures. Specifically, a categorical Recurrence Plot is often composed of rectangular structures rather than line structures (diagonal and horizontal/vertical), over which the recurrence quantification measures were originally proposed. Starting from this observation, I consider alternative computational procedures to extract a non-biased measure of entropy for the categorical case, showing the viability of such a choice with simulated data (C) 2018 Elsevier B.V. All rights reserved.