A random forest method for real-time price forecasting in New York electricity market

Abstract

This paper mainly focuses on the real-time price forecasting in New York electricity market through random forest. Accurate forecasting is regarded as the most practical way to win power bid in today's highly competitive electricity market. Comparing with existing price forecasting methods, random forest, as a newly introduced method, will provide a price probability distribution, which will allow the users to estimate the risks of their bidding strategy and also making the results helpful for later industrial using. Furthermore, the model can adjust to the latest forecasting condition, i.e. the latest climatic, seasonal and market condition, by updating the random forest parameters with new observations. This adaptability avoids the model failure in a climatic or economic condition different from the training set. A case study in New York HUD VL area is presented to evaluate the proposed model.