Forecasting is a key feature in the analysis of statistical data. It entails estimating the current state using observations and the prior knowledge of the state. The nonlinear and non stationary nature of the electricity data calls for a better forecasting method. In this study the States space models under Kalman filter method and Ensemble Kalman filter are used. This is a variance minimizing where for each step it minimizes the variance of the estimation errors resulting to an optimal estimate. Having an ensemble forecast is of interest in the study to check how effective it is compared to a single forecast. This paper also gives the mathematical approach for each method. Electricity price data for a five year period obtained from Nordpool (United Kingdom set) was used.
Keywords: Forecasting, States space model, Kalman filter (KL), Ensemble Kalman filter (EnKF), Ensembles.
[ Download ]
