A Retrieval-Augmented Generation (RAG) Chatbot that provides not just raw weather data, but actionable, human-like advice. It combines live weather updates with semantic search and a large language model to help users make better daily decisions — like whether to carry an umbrella, plan outdoor activities, or adjust travel schedules.
- 🌍 Live Weather Data – Retrieves real-time weather conditions using OpenWeather API.
- 🔍 Semantic Search with FAISS – Efficiently stores and retrieves relevant weather knowledge.
- 🤖 AI-Powered Responses – Explains forecasts in natural, human-like language using Gemini 2.5 Flash.
- 📊 Actionable Insights – Answers practical queries like “Should I carry an umbrella tomorrow?”.
- 💻 User-Friendly UI – Interactive chatbot built with Gradio.
- Data Source – OpenWeather API
- Embedding Model – All-MiniLM-L6-v2
- Vector Database – FAISS
- LLM (AI Model) – Gemini 2.5 Flash
- Chatbot UI – Gradio
- User submits a query (e.g., “Should I carry an umbrella tomorrow in Hyderabad?”).
- Weather data is fetched in real-time from the OpenWeather API.
- Query + weather info are embedded using MiniLM.
- FAISS retrieves the most relevant vectors.
- Gemini 2.5 Flash generates a simple, context-aware recommendation.
- Response is displayed instantly in the Gradio UI.
✅ Example:
User: “Should I carry an umbrella tomorrow?”
Bot: “Given the light to moderate rain and significant rainfall (29.9mm), it is recommended to carry an umbrella or a raincoat.”
The system integrates live weather APIs, embeddings, vector retrieval, and an LLM for contextual reasoning.
- Multi-day forecast support with detailed advice
- Personalized alerts (commute, farming, events, travel)
- Deployment with FastAPI or Streamlit for web access
- Integration with voice assistants for hands-free predictions
- Clone the repository:
gh repo clone Pavan47ML/Weather_Prediction_using_RAG cd weather-rag-chatbot - Run the code