Predictive models are the backbone of optimizing urban management for sustainability in smart cities. Therefore, this work stresses Energy Load Forecasting for Renewable Energy Management while highlighting real-time predictive models to balance supply with demand in a power network at different temporal levels with integrated renewable sources. As these renewable sources have variabilities, accurate forecasting has emerged as crucial for grid stability and optimizing resource distribution. This paper aims to study in detail the machine learning algorithms, combined with real-time data obtained through IoT devices, that enable effective energy management strategies. In particular, this work focuses on: These involve comparative performances of various prediction algorithms concerning accuracy, scalability, and responsiveness in real-time applications, including pointing out how models will contribute significantly to big data streaming with adequate integrations of devices to the IoT and cloud infrastructures. While predictive analytics is of great benefit, there remain challenges with data quality, model scalability, and the dynamic nature of urban energy systems. In addressing these challenges and proposing novel solutions, this work contributes to exploring how predictive analytics can help improve smarter city management and facilitate sustainable energy practice.