Comparative Analysis of Regression Models for Household Appliance Energy Consumption Prediction using Extreme Gradient Boosting

This research paper presents a comprehensive study on the prediction of energy consumption for household appliances using machine learning algorithms, with a focus on regression models. The performance of five popular regression models, namely XGB Regressor, Linear Regressor, Ridge Regressor, Support Vector Regressor (SVR), and K Neighbors Regressor, is evaluated. The experiment is conducted using a household appliances dataset from Belgium. Evaluation metrics, including mean absolute error (MAE), mean absolute percentage error (MAPE), and root mean square error (RMSE), are used to assess the performance of each model and facilitate a thorough comparison. The results obtained reveal that the XGB Regressor model outperforms the other models in terms of MAE, MAPE, and RMSE, with significantly lower values indicating superior predictive accuracy. The XGB Regressor achieved an MAE of 0.322, MAPE of 1.338, and RMSE of 0.699, outperforming all other models. The findings highlight the effectiveness of the XGB Regressor in accurately predicting energy consumption for household appliances. Additionally, the Linear Regressor, Ridge Regressor, SVR, and K Neighbors Regressor also demonstrate reasonably good performance, albeit with slightly higher errors compared to the XGB Regressor. Through this research, accurate energy consumption prediction models for household appliances are developed, contributing to the advancement of energy efficiency in the residential sector and the comparison of various regression models, with the XGB Regressor, helps stakeholders be empowered to make data-driven decisions, refine energy management approaches, and champion sustainable energy initiatives.