A Machine Learning Approach to Residential Energy Prediction Using Large-Scale Datasets in MongoDB

This paper investigates the efficacy of machine learning algorithms for forecasting energy consumption within residential environments. A comprehensive dataset containing over 2.07 million observations of individual household electric power consumption, housed within a MongoDB database, serves as the foundation for this analysis. Three well-established regression models, Linear Regression, Gradient Boosting, and XGBoost, are evaluated. Their performance is assessed using Mean Absolute Error (MAE), Mean Squared Error (MSE), and R-squared metrics. The primary objective is to identify the optimal approach for household energy management by evaluating the effectiveness of each model in accurately predicting energy consumption patterns. By harnessing the predictive capabilities of machine learning and MongoDB's efficient data management functionalities, this research reveals that Gradient Boosting emerged as the best-performing model. This finding suggests that Gradient Boosting offers valuable insights for optimizing energy consumption and promoting sustainable practices. This work contributes to the expanding body of knowledge concerning the application of machine learning for energy prediction, particularly within the context of readily scalable databases like MongoDB. The identified superiority of Gradient Boosting, applied to a large dataset, has the potential to serve as a cornerstone for the development of intelligent energy management systems, ultimately promoting efficient and sustainable household energy consumption.