Electric Load Forecasting & Congestion Modeling 

In this project, we developed machine learning models to forecast electricity demand and predict transmission congestion across 9 regions of New York State, using data from the New York Independent System Operator (NYISO). We focused on modeling real-time energy prices, demand loads, and weather conditions (temperature and humidity) between 2021–2024.

Using XGBoost regression, we achieved high accuracy in forecasting hourly electricity demand, with R² scores as high as 0.97 (NYC and Long Island) and Mean Absolute Percentage Errors (MAPE) between 2–5% across all regions. We also trained XGBoost classifiers to detect congestion zones, achieving up to 98% accuracy in certain regions.

This project demonstrated the power of gradient boosting algorithms in modeling complex, nonlinear relationships in utility data, and offered practical insights into optimizing energy grid performance and identifying congestion risk factors such as seasonality, weather, and regional infrastructure variability.

Tools:Python, XGBoost, Scikit-learn, Pandas, Matplotlib  

Key Skills:Regression, Classification, Model Evaluation, Data Cleaning, Feature Engineering