A novel approach for converting spatio-temporal series into complex networks

TL;DR

This paper introduces the Gravitational Graph algorithm, a physics-inspired method for transforming spatio-temporal data into complex networks, demonstrated on air pollution data to reveal hidden relationships.

Contribution

The study presents a novel physics-based approach for converting spatio-temporal series into networks, extending beyond traditional time series transformations.

Findings

The GG algorithm effectively captures spatial and temporal relationships.

Application to air quality data uncovers hidden dependencies.

The method enables analysis of complex environmental systems.

Abstract

This study aims to offer a new perspective on complex network representation of real-world systems. Currently, the most well-known transformation algorithms in the literature treat each data point in a time series as a node and transform the time series into a network. In this study, we present a new approach converting spatio-temporal series into a complex network. We focus on studying this transformation by grounding it in the context of physics , with the aim of adapting it to real-world problems, which often manifest as complex systems across various domains. We introduce the Gravitational Graph (GG) algorithm, which is grounded in the concept of gravitational force from fundamental physics. We consider air pollution concentrations, which represent a global environmental health risk, as an example of a complex environmental system, and apply the GG algorithm to particulate matter of 10 microns (PM10) recorded by 21 air quality monitoring stations located in various regions of Istanbul, Turkiye. While the GG algorithm allows for the conversion of spatio-temporal series -- rather than time series -- into networks, it also enables the analysis of the statistical properties and characteristics of the converted networks, thereby uncovering hidden relationships and dependencies that may not be apparent in the original time series.