Learning Predictive Leading Indicators for Forecasting Time Series Systems with Unknown Clusters of Forecast Tasks

TL;DR

This paper introduces a novel multi-task learning approach for forecasting multiple interrelated time series by discovering leading indicators and cluster structures, improving accuracy over traditional VAR models.

Contribution

The method jointly learns forecast models and identifies leading indicators within the system, linking Granger causality to sparse graph inference in a multi-task framework.

Findings

Outperforms baseline VAR models in synthetic and real data.

Effectively discovers leading indicators and task clusters.

Enhances forecast accuracy through shared information.

Abstract

We present a new method for forecasting systems of multiple interrelated time series. The method learns the forecast models together with discovering leading indicators from within the system that serve as good predictors improving the forecast accuracy and a cluster structure of the predictive tasks around these. The method is based on the classical linear vector autoregressive model (VAR) and links the discovery of the leading indicators to inferring sparse graphs of Granger causality. We formulate a new constrained optimisation problem to promote the desired sparse structures across the models and the sharing of information amongst the learning tasks in a multi-task manner. We propose an algorithm for solving the problem and document on a battery of synthetic and real-data experiments the advantages of our new method over baseline VAR models as well as the state-of-the-art sparse VAR learning methods.