Electric power price forecasting using data mining techniques

Abstract

Electricity price is the governing factor in taking various operational decisions such as generation scheduling, exchange of power amongst utilities, trading of power in market along with keeping pace with technical stability and reliability of power system. The accurate forecasting of price of electric power is a need of every participant in restructured power system scenario. Hence, this paper is an attempt to apply data mining for forecasting the electricity price. The k-mean algorithm is used for classification of data of historical prices of New York Energy Market (NYISO) according to type of day, into three classes. The k-NN algorithm to divide the classified data into two patterns for month of February-March and April to January. Once classification is done, the data is used for developing forecasting model. The historical electricity price data of 2014, along with load is used as input patterns. The accuracy of the developed model is verified by forecasting respective period samples of 2015. The performance of forecasting model is very satisfactory. The step wise development of forecasting model and the results are discussed in detail.