Time-series Scenario Forecasting

TL;DR

This paper introduces a Bayesian dictionary learning approach for time-series scenario forecasting, enabling probabilistic sampling of entire future sequences, and demonstrates its effectiveness in temperature prediction for Houston, especially where physics-based models are unavailable.

Contribution

The paper presents a novel Bayesian dictionary learning method for generating probabilistic time-series scenarios, addressing limitations of traditional point-wise error bars.

Findings

Performs comparably to physics-based ensemble methods for temperature forecasting.

Enables sampling from the posterior distribution of entire time-series.

Shows promise for applications lacking physics-based models.

Abstract

Many applications require the ability to judge uncertainty of time-series forecasts. Uncertainty is often specified as point-wise error bars around a mean or median forecast. Due to temporal dependencies, such a method obscures some information. We would ideally have a way to query the posterior probability of the entire time-series given the predictive variables, or at a minimum, be able to draw samples from this distribution. We use a Bayesian dictionary learning algorithm to statistically generate an ensemble of forecasts. We show that the algorithm performs as well as a physics-based ensemble method for temperature forecasts for Houston. We conclude that the method shows promise for scenario forecasting where physics-based methods are absent.