Econometrics as Sorcery

TL;DR

This paper introduces a graph-theoretic clustering method for identifying dependencies among multiple time series using linear models and energy-based similarity measures, generalizing existing multivariate procedures.

Contribution

It proposes a novel dynamical clustering approach that extends traditional multivariate methods by incorporating energy-based similarity and graph theory.

Findings

Effective grouping of processes based on model energy minimization

Generalization of existing multivariate clustering techniques

Provides a new framework for analyzing dependencies among time series

Abstract

The paper deals with the problem of identifying the internal dependencies and similarities among a large number of random processes. Linear models are considered to describe the relations among the time series and the energy associated to the corresponding modeling error is the criterion adopted to quantify their similarities. Such an approach is interpreted in terms of graph theory suggesting a natural way to group processes together when one provides the best model to explain the other. Moreover, the clustering technique introduced in this paper will turn out to be the dynamical generalization of other multivariate procedures described in literature.