A general method for electricity market price spike analysis

Many techniques have been employed to forecast electricity market prices, and have achieved good results. These techniques mostly focus on normal price forecasting, not on the forecasting of price spikes. Data mining techniques have been successfully applied to forecast the value of price spikes under the condition that a spike appears. However, an effective method of predicting the occurrence of the spikes is yet to be seen. In this paper, a data mining based approach is presented to give a reliable forecast of the occurrence of price spikes. Combined with spike value prediction techniques, the proposed approach can give a comprehensive price spike forecasting. In this paper, data pre-process techniques are described to find the attributes relevant to the spikes. Then a simple introduction to the classification techniques is given for completeness. Two algorithms: support vector machine and probability classifier are chosen and discussed in detail. Actual market data are used to test the proposed model, and promising results have been obtained.