If you are interested in machne learning you might check out the American Meteorological Society Wind Power Prediction Contest. It is sponsored by the AMS Committee on AI Applications to Environmental Science, Energy, and Probability and Statistics. The goal is to compare various statistical learning techniques to predict wind power production. This contest considers a wind farm consisting of 53 turbines located in northern Colorado. The objective is to forecast the wind power for the wind farm (sum of the power from the 53 turbines). The inputs are forecast variables from several Numerical Weather Prediction (NWP) models, both from the National Center for Environmental Prediction and NCAR's high resolution WRF forecast. Training data are currently available on the download site, and the test data will be available in November.
The contest consists of two parts: (1) a day-ahead forecasting contest and (2) a short-term (one-, two-, and three-hour) contest. Participants are welcome to enter either or both parts of the contests. The AMS welcome entries from professionals to students, but registration is required by November 15. Entrants names will be kept blind – they must register an abstract to present a poster at AMS, then winners will be notified of an invitation to present a talk at the January AMS Annual Meeting in Atlanta. The AMS expects to solicit prize money, which will be announced on the website at a later time.