vshar102/Morocco-SmartGrid-AI-Driven-Energy-Forecasting-Renewable-Integration

AI-powered project to forecast Morocco's energy consumption and optimize renewable energy integration into the grid. Combines machine learning, time-series analysis, and smart grid strategies to enhance sustainability and energy efficiency. # Morocco-SmartGrid-AI-Driven-Energy-Forecasting-Renewable-Integration Developed an AI-powered system to forecast Morocco's electricity demand and support the integration of renewable energy into the national grid. Leveraging a dataset of 61,755 consumption records across 9 key variables, the project utilized advanced time-series analysis and machine learning techniques to enhance sustainability and energy efficiency. Eight forecasting models—including ARIMA, SARIMA, Random Forest, XGBoost, LSTM, and Prophet—were implemented and systematically compared to optimize load prediction accuracy for utility-scale deployment and smart grid planning. # Key Achievements: Reduced forecasting error by 10.9% using Random Forest model (RMSE: 1,922.07) compared to seasonal baseline methods, enabling more accurate grid load balancing and resource allocation decisions Outperformed 7 competing models through systematic performance evaluation, with Random Forest achieving superior accuracy over deep learning approaches (LSTM RMSE: 45,581) and traditional statistical methods Identified critical consumption drivers through feature importance analysis, revealing that lag-1 consumption patterns (85.5% importance) and 7-day rolling averages (9.5% importance) are primary predictors for energy demand forecasting Enhanced model interpretability by incorporating weather correlation analysis (temperature, humidity, wind speed) and engineered temporal features (seasonal decomposition, cyclical day-of-week patterns) for comprehensive demand understanding # Technical Implementation: Applied advanced feature engineering including lag variables, rolling window statistics, and cyclical time encoding to capture complex seasonal consumption patterns. Conducted rigorous model comparison framework across statistical, machine learning, and deep learning paradigms, establishing Random Forest as the optimal solution for production energy forecasting systems. # Business Impact: Delivered actionable ins …