Smart Grid Load Forecasting using Machine Learning
- Built a Random Forest model to forecast hourly electricity demand using synthetic smart grid data
- Used 21,900 hourly data points (2.5 years) with seasonality and noise on a 200 MW baseline
- Engineered key temporal features: hour, day, month, year, and holiday indicators
- Model Performance: MAE of 8.16 MW (4.1%) and RMSE of 10.23 MW (5.1%)
- Visualized predicted vs. actual loads with performance metrics for interpretability
- Practical, hands-on application of core ML skills: data simulation, feature engineering, model training, and evaluation
Technologies Used:
Python
scikit-learn
pandas
matplotlib
Machine Learning