Day-Ahead Electricity Demand Forecasting: Post-COVID Paradigm
Leaderboard (numbers are kW MAE):
Teams with more than 5 missing submissions are eliminated from the leaderboard.
- Data is now available to registered users.
- Registration closed on March 1, 2021.
- Submissions are also accepted by sending the csv file to firstname.lastname@example.org, using the same email subject as the file name.
- Evaluation period ended on April 14, 2021.
- Reporting period is ongoing.
- Final report should provide:
- The name, email, and affiliation of all the team members;
- Diagram and description of models used throughout the evaluation period (max 2 pages);
- The training and inference codes for all the models used throughout the evaluation period;
- A readme file explaining codes execution and dependencies.
The ongoing COVID-19 related shutdowns have had a profound impact on the electric demand profiles worldwide, as governments put strict mitigation and/or suppression measures in place. The global electrical demand plummeted around the planet in March, April, and May 2020, with countries such as Spain and Italy experiencing more than 20% decrease in their usual electric consumption. In view of such massive electric demand changes, electricity network operators are facing unprecedented challenges in scheduling energy resources, as energy forecasting systems struggle to provide an accurate demand prediction. In fact, power systems operational reliability highly depends on an accurate projection of the future demand and scheduling an appropriate mixture of generation resources accordingly. Particularly, day-ahead forecasts are critical in managing market operation uncertainty. Thus, recent changes expose operators to technical and financial risks, further reinforcing the adverse economic impacts of the pandemic.
This competition aims at a detailed analysis of the impacts of the COVID-19 related measures on electricity demand, calling for strategies to mitigate the impact on day-ahead forecasting techniques’ performance. In particular, the competition is focused on day-ahead prediction of city-wide demand. The competition includes one-track only, deterministic forecasting of hourly load, 16 to 40 hours ahead.
Data includes historical demand, historical weather observations, and historical weather forecasts. Data is obtained from a real-world utility and weather service providers, and thus may be contaminated with issues such as missing periods, anomalies, etc. Details of data will be provided to registered participants.
The test data will cover 30 batches of 1-day each. Evaluation takes place in 30 phases. Contestant must provide a day-ahead forecast based on the most recent released data. More details with respect to submission rules will be released closer to the evaluation period.
- Participants with 6 or more missed submissions will be disqualified from the competition.
- Missed submissions of remaining participants, if any, will be replaced by the competition naïve benchmark. The technical committee will use its discretion to make sure that missed submissions have minimal impact on the final ranking after the final submission.
- Participants will be ranked based on the MAE of the entire evaluation period.
- In the event of a tie (defined as statistically insignificant difference in MAE) the organisers reserve the right to decide final, potentially joint, placings at their discretion following analysis of forecast performance and receipt of documentation.
- Each day during the evaluation period, a new file will be uploaded between 12:30-1 PM EST, containing data up to 8 AM of same day.
- Submission for day-ahead forecasts of 16 to 40 hours ahead are allowed up to 11 AM EST of next day.
- Submissions should be in CSV format.
- CSV file name should be “Target Month-Target day-Last name-First Name”; for example, if the file being submitted contains predictions for March 21 (being submitted between March 20, 1 PM EST to March 21, 11 AM EST), the file name would be March-21-Farrokhabadi-Mostafa.csv
- The submitted CSV files should contain one single column of 24 float values; no header is accepted.
- Those submitting results as a group will be given the opportunity to include other members’ name in the final report and code submission.
- Multiple submissions under same name and/or email will be disqualified. Submission under emails other than those registered will be disqualified.
- Use of any exogenous input data, of any sort, except those already provided, will disqualify the participant.
- Submissions violating any of the above will be disqualified.
- The Technical Committee reserves the right to disqualify any submission deemed ineligible.
A naive benchmark is included for reference and to fill up to five missing entries from participants. The benchmark forecast for day D is the load from the most recent complete day of the same type available at the time of making the forecast (8AM on D-1). Day types are Saturday, Sunday and Weekday.
- Dec. 14, 2020: Historical data release
- Mar. 1, 2021: Registration deadline
- Mar. 15, 2021: Evaluation period starts – first batch of test data release
- Apr. 13, 2021: Evaluation period ends – last batch of test data release
- Apr. 19, 2021: Final report and code submission deadline
- May 3, 2021: Winners announced
- 1st place: $5000 USD
- 2nd place: $3500 USD
- 3rd place: $1500 USD
IEEE OAJPE is running a new call for papers for Special Section on “COVID-19 Impact on Electrical Grid Operation: Analysis and Mitigation”. The winners of the competition are pre-approved for submission in the SS. For more information, please visit the SS link.
- Registrants should use an official academic, government, or industry instituition email. Personal emails will not be accepted.
- The winners’ identity will be verified through official government-issued IDs before the announcement of winners.
- Dr. Jethro Browell, University of Strathclyde, Scotland
- Dr. Mostafa Farrokhabadi, BluWave-ai, Canada
- Dr. Stephen Makonin, Simon Fraser University, Canada
- Dr. Wencong Su, University of Michigan-Dearborn, US
- Dr. Yi Wang, ETH Zurich, Switzerland
- Dr. Hamidreza Zareipour, University of Calgary, Canada
- IEEE DataPort
- IEEE PES PSOPE WG on Energy Forecasting and Analytics
- IEEE Foundation Donor Supported Program
- Data is provided courtesy of BluWave-ai.
Note: The Technical Committee reserves the right to amend the rules and regulations without giving prior notification or any reasons thereof.
Please refer to the above Submission Rules.