Skip to content

Abdullah12389/Crypto_Assistant

Repository files navigation

📈 Bitcoin Price Predictor & Financial Chat Assistant

This project is an intelligent app that predicts the future price of Bitcoin using deep learning models like LSTM and Facebook Prophet. It features a real-time chatbot powered by LangGraph, giving personalized financial advice and search functionality across a local Bitcoin database and the internet.

🚀 Features

  • 🔮 Bitcoin Price Prediction using LSTM and Prophet models
  • 🤖 AI Chatbot for financial advice, news, and Bitcoin questions
  • 🌐 Web Search Integration for real-time financial data and news
  • 📊 Real-time Bitcoin Graph using Plotly
  • 🔁 LangGraph-powered Workflow to manage chatbot logic and data search
  • 🖼️ Visual LangGraph Structure shown below

🧠 Technologies Used

  • LangGraph — chatbot logic management
  • LangChain — LLM integration
  • LSTM — time-series deep learning model for price forecasting
  • Facebook Prophet — interpretable time-series forecasting
  • Plotly — real-time charting
  • Flask HTML CSS JAVASCRIPT — frontend interface
  • Pandas + Numpy — data processing

📥 Installation

git clone https://github.com/yourusername/bitcoin-predictor
cd bitcoin-predictor
pip install -r requirements.txt

⚙️ How It Works

  1. User inputs a question or query about Bitcoin
  2. LangGraph chatbot decides whether to:
    • Search internet
    • Query Bitcoin DB
    • Fetch financial news
    • Run full prediction pipeline
  3. Prediction is done using LSTM + Prophet
  4. Chatbot responds with a result
  5. Graph updates in real time

📊 Real-Time Bitcoin Graph

The GUI shows a live-updating Bitcoin candlestick chart using Plotly, updated via WebSocket or polling. It includes:

  • Price changes
  • Prediction overlays
  • Volume and RSI indicators (if added)

🔄 LangGraph Workflow

LangGraph Structure

✅ Future Improvements

  • Trading strategy simulations
  • Social media sentiment analysis
  • Telegram/Discord bot version

📄 License

This project is open-source under the MIT License.

Releases

No releases published

Packages

 
 
 

Contributors