Campus Energy forecast- ML powered dashboard for hostel energy optimization

University hostels and college buildings in India face skyrocketing electricity bills due to unpredictable usage patterns — especially during peak summer AC loads, overnight lights/fans, and mess/kitchen operations. Traditional monitoring is manual, reactive, and inefficient, leading to 20–40% avoidable wastage, overload penalties, and sustainability challenges.

Smart Hostel Energy Forecaster is a full-stack, web-based intelligent dashboard designed specifically for educational institutions. It uses hybrid machine learning models to deliver accurate short-term energy consumption forecasts , real-time appliance/zone-level monitoring, automatic anomaly detection and cost simulation with monthly limit alerts.

Key Features:

• Data Upload & Storage — Easy CSV upload of historical meter/sub-meter readings; stored in SQLite for quick querying.

• Predictive Forecasting — Time-series models trained on features like timestamp, zone (e.g., Wing A, Mess), occupancy simulation, weather proxies, and academic schedule dummies → outputs predicted power (Watts/kWh) visualized in interactive Chart.js graphs.

• Anomaly Alerts — Unsupervised ML (Isolation Forest) flags abnormal patterns (e.g., geyser running at 3 AM).

• Billing & Optimization Insights — Real-time cost calculation (₹/kWh input), projected monthly bill, and "what-if" simulations (e.g., shift loads to off-peak).

• User-Friendly Dashboard — Responsive HTML/CSS/JS frontend with live usage tables, toggleable forecasts per zone/appliance, and alert notifications.

Impact & Hackathon Fit:
Built for real-world Indian campuses, this project promotes green campus goals, reduces bills by 15–30% through early warnings & behavioral nudges, and demonstrates end-to-end ML deployment — from data ingestion to interactive UI. Fully open-source, extensible for IoT integration.