Submission · anonymous
This paper presents the results of a day-ahead electricity demand forecasting competition that was motivated by the unprecedented changes in electricity consumption patterns caused by the COVID-19 pandemic. Participants were challenged to forecast 24-hour-ahead electricity demand for a large North American utility using data spanning the COVID-19 period. The paper describes the competition design, the evaluation framework, and the methods employed by top-performing teams, highlighting that ensemble and machine-learning approaches outperformed classical statistical baselines.