Tuesday, August 28, 2012

More about GEFCom2012

Up to date, 100+ people from the following 31 countries have filled in the Call For Participants survey:
Australia, Belgium, Brazil, Canada, China, Croatia, Cyprus, Denmark, France, Germany, Greece, Hungary, India, Iran, Italy, Japan, Mexico, Netherlands, New Zealand, Norway, Pakistan, Poland, Russia, Singapore, South Africa, Spain, Sweden, Switzerland, Turkey, UK, US
Kaggle is setting up the online competition environment for us, here are some more details about the competition:

1. GEFCom is not a paper contest. Instead, this is a competition that requires participants to develop models and submit forecasts based on a given data set. Accuracy of the forecasts will be one evaluation criteria.
2. In addition to accuracy, the participants are also required to submit a report describing the methodology, findings and models. Selected entries will be invited to IEEE PES General Meeting 2013 at Vancouver, Canada to present their methodologies and results. The final winners will be determined by the GEFCom Award Committee after the presentations based on forecasting accuracy, clarity of documentation, rigors of the approach, interpretability of the models and practicality to the industry.
3. A few entries will be invited to submit the report in scientific paper format to prestigious scholarly journals, such as International Journal of Forecasting and IEEE Transactions on Smart Grid.
4. The competition will be hosted at
Kaggle.com. We do encourage you to register an account at Kaggle if you haven't done so and take a look at how other competitions were conducted in the past.
5. We plan to start the competition on 8/31/2012 and end it on 10/31/2012.
6. For both load forecasting and wind forecasting tracks, you will be doing forecasts for multiple series. You can join either track or both tracks. We expect one to spend 80+ hours to get some decent forecasts for each track.
7. You can collaborate with your co-works or friends to form teams to join the competition.
8. You can also join the competition on behave of your company.
(The information below is about the load forecasting track):
9. 20 load series at delivery point level and 11 temperature series for a US utility will be given at the load forecasting track.
10. In each load series, 8 nonconsecutive weeks were taken out from the history. For each of the 20 series, the participants are required to backcast the 8 weeks in the history and the one week immediately following the history period. In addition, the participants are required to backcast the 8 weeks in the history and forecast 1 week after the history period at the system level, which is the sum of the 20 series.
11. Temperature data is available throughout the history, but not available for the one week following the history period. The use of temperature data is optional. The participants can make their own decision.
12. Different weights will be given at different levels and weeks when evaluating accuracy: weight of the last week at system level > weight of a history week at system level > weight of the last week at delivery point level > weight of a history week at delivery point level.
13. In each hour, sum of delivery point level loads is required to be equal to the system level load. Submissions that violate this requirement will be disqualified.
14. Complete instructions will be provided through Kaggle and our competition webpage.
For more information, please contact Dr. Tao Hong (hongtao01@gmail.com) or visit www.gefcom.org.

No comments:

Post a Comment

Note that you may link to your LinkedIn profile if you choose Name/URL option.