SIH 2025 — Buoy Data SQL Agent

Ask questions about Indian Ocean moored-buoy data in plain English. A LangGraph agent turns your question into validated SQL, runs it against a 4-year INCOIS ocean dataset, and answers with charts.

Built for Smart India Hackathon 2025 — an AI agent that makes oceanographic data from India's moored buoy network (INCOIS) queryable by anyone, no SQL required.

---

What it does

The system scrapes 4 years of moored-buoy observations (air/water temperature, wind speed, humidity, rainfall, radiation, pressure) across multiple buoy stations (AD06, AD08, AD10, BD10–BD14), loads them into SQLite, and exposes a natural-language interface:

You:  "What was the average wind speed at BD11 in June 2023?"
Agent: SELECT AVG(wind_speed) FROM ... WHERE buoy='BD11' AND month=6 AND year=2023
       → 8.4 m/s 📈 (with a chart)

Architecture

┌─────────────┐   ┌──────────────────┐   ┌──────────────┐   ┌─────────────┐
│  Streamlit  │──▶│  LangGraph agent │──▶│  SQL (SQLite)│──▶│  INCOIS DB  │
│   / CLI     │   │  (validated SQL) │   │   execution  │   │  (4-yr data)│
└─────────────┘   └──────────────────┘   └──────────────┘   └─────────────┘
                       │  ▲
                       ▼  │  memory / chat history
                   ┌──────────────┐
                   │  LLM manager │  (OpenAI / OpenRouter)
                   └──────────────┘

• get_buoy_data.py — scraper for INCOIS moored-buoy stock data (per-parameter URLs, configurable year range)
• run_master_scraper.py — fetches the full buoy metadata list from the INCOIS API and orchestrates bulk scraping
• db_toCSV.py — exports the SQLite database to CSV
• agent/ — LangGraph workflow: SQLAgent (generate + validate SQL) → execute → DataFormatter (chart-ready output) → final answer
• app.py — Streamlit chat UI with conversation memory

Quick start

pip install -r requirements.txt   # or: pip install streamlit langgraph langchain-openai pandas

# 1. Set your LLM keys
cp .env.example .env              # fill in OPENAI_API_KEY or OPENROUTER_API_KEY

# 2. (Re)build the database from scraped data
python db_toCSV.py

# 3a. CLI
python main.py

# 3b. Streamlit dashboard
streamlit run app.py

Data

• Source: INCOIS moored buoy data (Indian National Centre for Ocean Information Services)
• 4-year dataset (SIH_4yr/), imported into data/sih_4y4.db
• Buoys: AD06, AD08, AD10, BD10, BD11, BD12, BD13, BD14
• Parameters: air & water temperature (multiple depths), air pressure, humidity, wind speed, rainfall, solar radiation

Security

API keys live in .env (git-ignored). .env.example documents the required variables.

License

MIT