Electricity contract price prediction using genetic programming with functional blocks

Abstract

The prediction of the prices of the contracts in non-regulated electric
energy markets is the key for the market agents to make strategic business and operational decisions. The average prices of the contracts sold in the Colombian electric energy market are predicted in this study by means of a modified genetic programming algorithm. The developed model is capable of capturing the intrinsic dynamics of the prices and the price predictions for the upcoming months with a more accurate precision than the ARIMA and DAN2 models for prediction horizons of 12 and 14 months, as they have been reported in the literature.