Prediction using ARFIMA and FOU models of affluent energy

Abstract

In this work we study predictions from ARFIMA and FOU models for the weekly data series of affluent energy generated by hydroelectric dams in Uruguay between 1909 and 2012. The estimation of Hurst coefficient suggests modeling through long memory time series. We present two families of time series models of this type, ARFIMA and FOU (fractional Ornstein-Uhlenbeck) models. Their parameters are estimated and taking into account their predictive power, their performance is compared.