Forecasting performance of time series and regression in modeling electricity load demand

Abstract

Electricity load demand modelling is considered as one of the important area among researchers since electricity are evolving throughout the time. There is a lot of technique to analyze the load demand such as by using classical method or conventional methods. However, most of the techniques only consider univariate data sets. The purpose of the current study is to evaluate the performance of time series and regression in load demand forecasting. Time series models are considered univariate data sets while regression model considered both univariate and multivariate data sets. Time series models considered in this study are Exponential Smoothing (ES) state space, Autoregressive Integrated Moving Average (ARIMA), Autoregressive Autoregressive (ARAR), and Autoregressive Moving Average Error, Trends and Seasonal Components (TBATS) while for regression model, Stepwise Multiple Regression will be considered by using Root Mean Square Error (RMSE) as a forecasting accuracy criteria, the study concludes that the Stepwise Multiple Regression method is more appropriate model.