Evaluation of Price Forecast Systems for Turkish Electric Market

Abstract

It is very important to forecast the electric prices in deregulated markets for both producers and brokers. This information is crucial to make effective decisions concerning to production, purchase, maintenance and investment. In this study, we built two different systems for short-term prediction of electricity price in Turkish Electric Market. One of the systems built on ARIMA model, while the other employs a feed forward neural network. Both systems use calendar and historical price information as input. Performance of both systems are compared and it is shown that it is possible to forecast weekly electric price with an average error rate of %8.5.