The AMS Solar Energy Prediction Contest brought together participants from around the world to tackle the problem of improving solar energy forecasting. The contest was held on Kaggle , a website for hosting worldwide data mining and machine learning contests. Participants used a set of NOAA Global Ensemble Forecast System reforecasts as input for statistical and machine learning models that predicted the total daily solar energy at 98 Oklahoma Mesonet sites. Over the four-month span of the contest, 157 teams from six continents submitted over 2,500 sets of predictions. The winners were:
First Place:
Lucas Eustaquio Gomes da Silva (Belo Horizonte, Brazil) and
Gilberto Titericz Jr. (São José dos Campos, Brazil)
Second Place:
Benjamin Lazorthes (Toulouse, France)
Third Place:
Owen Zhang (New York, New York)
Top Student Winner:
Gilles Louppe (Liège, Belgium)
The first-place, second-place, and student winners will present their methods at a special session (Wednesday, February 5, 1:30-2:30 p.m., Room C204) of the AMS Annual Meeting. Specialists in renewable energy and data science as well as all interested attendees are invited to learn about the methods used in the contest and to discuss what value the contest results may provide for forecasts of renewable energy and other phenomena.
The contest is jointly sponsored by the 12th Conference on Artificial and Computational Intelligence and its Applications to the Environmental Sciences; the 22nd Conference on Probability and Statistics in the Atmospheric Sciences; and the Fifth Conference on Weather, Climate, and the New Energy Economy.
You can find more information about the contest here. The winners’ model approaches and codes are available here.