Long-Term Load Forecasting At The End Of The Year Using The Linear Regression Method

This study discusses long-term electricity load forecasting at the end of the year using a linear regression method, applied to the Glugur Substation in Medan. Historical load data from 2021 to 2024 is used as the basis for calculations to forecast electricity loads until the end of 2026. The forecasting process is carried out by modeling the relationship between the increase in time to peak load and monthly energy, using a linear regression approach. The forecasting results show a stable and consistent load increase trend. Model accuracy evaluation is carried out using MAPE, R², MAE, and RMSE metrics. The lowest MAPE value is obtained at 1.37% at the Selayang Substation, while the Glugur Substation has a MAPE value of 2.43%, indicating a good level of accuracy for long-term forecasting. This research contributes to supporting the planning of a more reliable and efficient electricity distribution system.