Abstract
This study addresses the critical research domain of electricity consumption prediction, emphasizing its importance in energy production, distribution, and related aspects such as load balancing, cost optimization, energy efficiency, and carbon emissions reduction. Various models have been explored to tackle the challenges of prediction accuracy. The research introduces a Temporal Convolution Network (TCN) as a base model, aiming to enhance accuracy in predicting electricity consumption using a United Kingdom (UK) dataset spanning 2009 to 2023. Models like ARIMA, Linear Regression (LR), Random Forest (RF), Support Vector Regression (SVR), LR-SVR Hybrid, and Long Short-Term Memory (LSTM) were compared using Mean Absolute Error (MAE), with the proposed TCN demonstrating superior accuracy over other models