An efficient non-recursive algorithm for transforming time series to visibility graph

In recent years, transforming a time series into visibility network has emerged as a powerful tool of data analysis, with applications in many pure and applied domains of statistical physics and non-linear dynamics. The algorithms available for this transform are either very slow or consume copious...

Full description

Saved in:

Bibliographic Details
Published in:	Physica A Vol. 514; pp. 189 - 202
Main Authors:	Ghosh, Saptorshi, Dutta, Amlan
Format:	Journal Article
Language:	English
Published:	Elsevier B.V 15.01.2019
Subjects:	Complex network Iteration and recursion Time series analysis Visibility graph algorithm Visibility graph algorithm Time series analysis Iteration and recursion Complex network
ISSN:	0378-4371, 1873-2119
Online Access:	Get full text
Tags:	Add Tag No Tags, Be the first to tag this record!

Description
Summary:	In recent years, transforming a time series into visibility network has emerged as a powerful tool of data analysis, with applications in many pure and applied domains of statistical physics and non-linear dynamics. The algorithms available for this transform are either very slow or consume copious amount of memory resorting to recursive calls. Here we propose an efficient non-recursive algorithm for constructing natural visibility graph from time series data. In comparison to the recursive method, the new algorithm offers safer and more optimized use of memory space without sacrificing its speed. Performance of this algorithm is tested with a variety of synthetic and experimental time series data-sets. •A novel sort-and-conquer algorithm is proposed to transform a time series into visibility graph.•Being non-recursive, this algorithm does not use stack frames in memory.•Speed of execution is either comparable or better than the existing divide-and-conquer algorithm.•Memory efficiency is significantly superior to the recursive method.
ISSN:	0378-4371 1873-2119
DOI:	10.1016/j.physa.2018.09.053