Machine Learning-Based Forecasting of Electricity Demand for Sustainable Electricity Planning

In particular, SDG 7 for universal access to sustainable energy and SDG 13 for climate action depend significantly on the amount of electricity consumed. Accuracy and scalability issues with traditional methods for predicting power usage present obstacles to achieving these SDGs. The effective use of historical data to predict energy consumption patterns and improve energy distribution and management is made possible by recent advancements in machine learning and data analytics. This paper proposed a model for forecasting electricity demand that utilized and compared several widely used machine learning algorithms. These models are Autoregressive Integrated Moving Average (SARIMAX), eXtreme Gradient Boosting (XGBoost), Random Forest, and Support Vector Regression (SVR), A historical time series benchmark dataset is adopted. XGBoost is the best machine learning algorithm for the adopted dataset, according to the experimental results. Moreover, the result revealed that the proposed model not only results in more efficient resource allocation and environmental outcomes, but it also significantly contributes to the broader sustainability goals outlined in the United Nations’ 2030 Agenda.