An interactive web app that predicts the likelihood of seal presence at any ocean location using real environmental data.

Click anywhere on the ocean to drop a pin, pick a date, and get an instant prediction powered by a machine learning model trained on thousands of real seal sightings. The app returns:

• Probability of seal presence at that location
• Top predicted seal species
• Real environmental conditions (ocean depth, sea surface temperature, wind speed, distance to shore)
• A fun seal fact 🦭

• Frontend: React, Mapbox GL JS, Vite
• Backend: FastAPI (Python)
• Model: Two-stage Logistic Regression trained on OBIS global seal occurrence data

• Node.js
• Python 3.10+
• A Mapbox public token (free at mapbox.com)

cd frontend
npm install
npm run dev

Create frontend/.env: VITE_MAPBOX_TOKEN=your_mapbox_token_here VITE_API_BASE=http://localhost:8000

cd backend
pip install -r requirements.txt
python -m uvicorn main:app --reload

Arun Malani, Ryan Tapia, Emily Hames, Minnie Kay