PREDICTION OF ELECTRIC ENERGY CONSUMPTION USING MACHINE LEARNING APPROACH FOR ETHIOPIAN ELECTRIC POWER AUTHORTIY WOLAITA DISTRICT

Abstract

Due to the increasing population, industrialization, and extensive use of technology, there has been a steady increase in the demand for electric energy. Accurate prediction of electric energy consumption plays a crucial role in optimizing power generation, distribution, and ensuring a stable power supply. However, there is a lack of study conducted on prediction of electric energy consumption in Wolaita Zone. The traditional methods used by previous researchers for forecasting are often limited in their ability to capture the complex patterns of electric energy consumption. In this study, researchers aimed to overcome these limitations by employing machine-learning algorithms to analyze historical energy consumption data and extract meaningful and useful patterns and trends. To achieve this objective the mixed research approach (both qualitative and quantitative) and experimental research design where employed. By achieving these objectives, this study will contribute to the advancement of energy consumption forecasting techniques and enable the stakeholders to make an informed decisions regarding energy generation, distribution, and pricing. The regression based machine learning-algorithm such as, Random forest; Extreme Gradient Boosting (XGBOOST), RNN, LSTM and GRU were employed for prediction of electric energy consumption. Based on the experiment-conducted recurrent neural network model suitable and outperforms others with R-squared 0.712, MAE of 0.021, MAPE 0.678% and MSE of 0.243.