Assessment of seasonal effects on city based daily electricity load forecasting using linear regression

Due to the importance of electricity and its impacts on the human living environment, several studies have been conducted to forecast and possibly reduce forecast errors of the electricity load. In this study, the load data of Iraq Sulaymaniyah city are used. Forward selection, backward elimination, and stepwise approaches are used to determine the variables in the multiple regression equation. The 6-year data of 2014-2019 was used to develop the day-ahead forecasting models, while the 2019 year was used as a test dataset to validate the model. Each year was divided into two-time intervals according to the change in load behaviour. The long-term seasonality effect was tried to be determined. The results show that instead of all data, the series divided into two according to long-term seasonality could estimate the load with lower errors. Divided series will help the control center to have a better estimation of electricity demand and energy purchase. Using our model, they will be capable of forecasting electricity load for upcoming months and years to replace the traditional way of calculating and reporting load. Keywords: Multiple Linear Regression, STLF, Load Forecasting, Control Center Component.