Energy Consumption Forecasting in Smart Cities Using Machine Learning
Prathamesh Anil Gurav ,
JVM's Degree College, Airoli.
Abstract :
Accurate forecasting of electricity consumption is a foundational requirement for smart cities, enabling grid stability, demand response, energy efficiency initiatives, and long-term infrastructure planning. Over the past decade, significant advances in data availability from smart meters and Internet of Things (IoT) devices, combined with progress in machine learning (ML) and deep learning (DL), have transformed the landscape of urban energy forecasting. This paper presents a comprehensive survey of energy consumption forecasting techniques applied in smart city contexts. We systematically review classical statistical approaches, machine learning models, deep learning architectures, hybrid techniques, and emerging transformer-based methods. The survey also examines commonly used datasets, feature engineering practices, evaluation metrics, uncertainty quantification techniques, and deployment considerations such as privacy, explainability, and federated learning. By synthesizing recent literature, this paper highlights strengths, limitations, and open research challenges, and provides guidance for researchers and practitioners in selecting appropriate forecasting models for different smart city scenarios.
