Pulse — Energy Consumption Forecasting

Energy consumption forecasting model using Random Forest and XGBoost regression — with feature engineering incorporating location-specific holidays and historical extreme weather events to explain anomalies and improve accuracy.

A first end-to-end machine learning project tackling energy demand forecasting through the full data science workflow — cleaning, EDA, and feature engineering before modelling. Key features engineered include location-specific public holidays and past extreme weather events to account for consumption anomalies that a baseline model would otherwise misread. A Random Forest and XGBoost regression ensemble was trained and evaluated, with results visualised in an interactive Streamlit and Plotly dashboard.