Abstract

  Univariate modelling is employed in this study to forecast the electricity consumption of the residential sector under the responsibility of the metropolitan electricity authority (MEA) over the five-year period from 2021 to 2025 based on time series data and by examining and comparing the effectiveness of three forecasting models. The three models under study are the Holt-Winters’ exponential smoothing, Brown exponential smoothing and Damped trend exponential smoothing. The Mean Absolute Percentage Error (MAPE = 2.35) is used to determine the most suitable and effective model, and the comparative results reveal the Dampedtrend model to be the most suitable for forecasting electricity consumption in the residential sector. Total electricity consumption from 2021 to 2025 for the residential sector is projected to continuously increase. The annual residential electricity capacity of the MEA shows a continuous decrease when compared to total electricity consumption between 2021 and 2025, with the distributed volume being 1,319.06, 1,004.56, 690.16, 375.85, and 61.64 million kWh respectively, due to the continuous decrease in electricity sales significantly impacting on the management of electricity production and distribution. Consequently, in the future, the MEA may not have sufficient electricity capacity to meet the consumption demand of the residential sector.

Keywords:Residential sector, Forecasting electricity consumption, Univariate modelling, Metropolitan