AI Data Analysis Assistant

A Streamlit application that uses OpenAI to generate and execute pandas code for data analysis. Upload a CSV file, ask questions in natural language, and get instant analysis with visualizations.

Features

• 📁 CSV File Upload - Easy file upload interface
• 👀 Data Preview - View first 10 rows and data statistics
• 💬 Natural Language Queries - Ask questions like "What are the top 5 products by revenue?"
• 🤖 AI-Powered Code Generation - Uses OpenAI GPT-4 to generate pandas code
• ⚙️ Safe Code Execution - Executes generated code in a restricted environment
• 📊 Multiple Output Formats - Results displayed as tables, charts, or text
• 💻 Code Learning - View generated code to learn pandas techniques
• 📚 Code History - Keep track of previous analyses

Requirements

• Python 3.11+ (or Python 3.12)
• OpenAI API key
• Streamlit
• Pandas
• Plotly

Installation

1. Clone or download this repository

2. Install dependencies:

   pip install -r requirements.txt

3. Set up your OpenAI API key:

   Create a .env file in the project root:

   OPENAI_API_KEY=your_api_key_here
   

   Or set it as an environment variable:

   export OPENAI_API_KEY=your_api_key_here

Usage

1. Run the Streamlit app:

   streamlit run app.py

2. Upload a CSV file using the file uploader

3. Review the data preview to understand your dataset

4. Ask questions about your data, for example:

   • "What are the top 5 products by revenue?"
   • "Show me a trend chart of sales over time"
   • "What is the average value by category?"
   • "Create a bar chart of total sales by region"

5. View results - The app will:

   • Generate pandas code to answer your question
   • Execute the code safely
   • Display results (tables, charts, or text)
   • Show the generated code for learning

Example Questions

The app can handle various types of data analysis questions:

• Aggregations: "What is the total revenue by category?"
• Rankings: "Show me the top 10 customers by purchase amount"
• Time Series: "Create a line chart showing sales trends over time"
• Comparisons: "Compare average sales across different regions"
• Filtering: "What are the products with sales above $1000?"

Safety Features

• Code execution is restricted to safe pandas and plotly operations
• Dangerous built-in functions are blocked
• Only allowed imports (pandas, plotly) are available
• Errors are caught and displayed safely

Project Structure

.
├── app.py              # Main Streamlit application
├── requirements.txt    # Python dependencies
├── README.md          # This file
├── .gitignore         # Git ignore rules
├── .env.example       # Environment variable template
└── LICENSE            # MIT License

Notes

• The app uses GPT-3.5-turbo for code generation. Make sure you have API credits available.
• Large CSV files may take longer to process.
• The generated code is displayed so you can learn and modify it if needed.

License

MIT License - see LICENSE file for details.