Studying connectivity between time-series using an interactive application

Abstract

[EN] Connectivity is a complex concept whose meaning varies for each domain. In a broad sense, it is possible to divide the connectivity in two main groups: physical and statistical. The former represents whether there is a physical link connecting the elements, like the telephone line, while the latter indicates a relation between the dynamics of the elements, as the number of sales and the income of a company. In several scientific fields, including economics, physics and neuroscience, the statistical connectivity is computed based on the analysis of time-series. Several different methods can be applied on the data, using different features of the signals, like covariance and frequency, to estimate the connectivity, yielding different results and interpretations. The covariance indicates if two elements follow the same dynamics, i.e., if they increase or decrease at the same time. On the other hand, the frequency is related to the time required for the elements to change, which can be daily in the case of the tides, or milliseconds if referred to the computations in a microchip. A third factor is the directionality, with one element depending on the other, but not in the opposite case. This work proposes an interactive computer-based application to generate a three-node network and estimate its connectivity following different approaches. We describe how the application is structured, from the selection of the parameters to the interpretation of the output results, detailing the skills in connectivity that can be developed by the students.