Advanced demand forecasting for strategic peak identification in Ontario’s electricity market

Marcelo Guarido, Kristopher A. Innanen

Forecasting market (energy) demand is an important task for large energy consumers and government agencies. In Ontario, IESO plays a crucial role in managing the electricity market by regulating different levels of energy payments for consumers. Large energy consumers, the Class A consumers, are particularly affected by these regulations as they are subject to time-of-use pricing, proportional to their consumption during the top 5 days with the highest provincial demand in the previous year. In this report, we proposed a two-step approach to forecast the top 5 peak days in Ontario's electricity market. The first step is to forecast the daily maximum market demand using Neural Networks, and the second step is to classify the top 5 peak days using XGBoost. The models were trained on data from January 1st, 2018, to December 31st, 2023, and tested on data from 2024. The forecasting model achieved an R2 score of 0.902 on the test set, and the classification model correctly predicted 3 of the 5 peak days, with the possibility to also analyze the probabilities for a day to be one of the top 5 peaks.