MACHINE LEARNING-ENABLED ENERGY FORECASTING USING IOT DATA FOR SMART GRID LOAD OPTIMIZATION

The integration of smart grids with Internet of Things (IoT) technology has transformed energy management systems by enabling realtime monitoring and efficient control of power distribution. As IoT-based energy meters become increasingly prevalent, accurately forecasting energy consumption has become critical for optimizing load balancing, minimizing energy waste, and enhancing demand response mechanisms. However, predicting energy usage is challenging due to varying consumer behaviors, environmental fluctuations, and seasonal trends. Traditional forecasting methods such as ARIMA and moving averages are limited by their inability to handle non-linear patterns, adapt to largescale data, or incorporate diverse features like weather data and time-based variables. These methods often struggle with noise, outliers, and dynamic consumption trends, resulting in poor prediction accuracy. Therefore, there is a growing need for intelligent, scalable, and adaptive systems that can analyze highdimensional data generated by smart meters. Addressing these limitations, the proposed system introduces a machine learning-based web application that uses Random Forest Regressor and Support Vector Regression (SVR) to forecast energy consumption. The system covers the complete pipeline, including preprocessing, outlier handling, feature selection, exploratory data analysis, model training, and evaluation, all integrated into an interactive Flask-based web interface. It efficiently captures the non-linear dependencies between various factors such as temperature, humidity, time, and calendar events with energy usage patterns. The use of advanced machine learning models improves forecasting accuracy, enabling power providers and stakeholders to make data-driven decisions, ensure grid stability, and support the development of sustainable energy practices in smart cities. By overcoming the limitations of traditional methods and effectively leveraging the richness of IoT-generated data, the proposed solution represents a significant step toward intelligent energy forecasting in nextgeneration smart grids