Forecasting energy consumption

This project applied time-series forecasting techniques to real-world loads in building facilities and factories.

The solution compared methods from exponential smoothing to gradient boosting and recurrent neural networks, with attention to edge-device performance.