Forecasting electricity consumption plays a crucial role in formulating effective energy policy. This ensures sustainable economic growth and supports national development. Accurate forecasting facilitates optimal power generation planning. It provides efficient resource allocation and reduces energy waste. Underestimating or overestimating energy demand can have significant consequences, such as energy shortages, forced outages, financial losses, and increases in electricity prices. This study introduces a hybrid model approach combining both eXtreme Gradient Boosting (XGBoost) and Multi-Layer Perceptron (MLP) to enhance electricity demand prediction. The proposed model is compared with Polynomial Regression, standalone XGBoost, MLP, and Long Short-Term Memory (LSTM) networks, utilizing four evaluation metrics: Mean Absolute Error (MAE), Mean Squared Error (MSE), Root Mean Squared Error (RMSE), and R2 score. The experiment results demonstrate that the hybrid model outperforms others, achieving an R2 score of 94.02% alongside low MSE (0.0007) and RMSE (0.0272). This indicates the model's superior accuracy and reliability on forecasting energy consumption. Future work aims to explore advanced methodologies, including transformer-based architectures, to further improve forecasting capabilities and address the challenges in electricity consumption prediction for diverse applications.