AI-Agent-with-Custom-MCP--For-Data-Analysis

# AI Agent for Data Analysis A modular, extensible AI-powered data analysis tool built with Flask, OpenAI, and a custom MCP server. Upload CSV files, ask questions, and get automated insights, visualizations, and summaries. --- ## Features - **CSV Upload:** Easily upload your dataset for analysis. - **Dashboard UI:** Modern, responsive interface for analysis and results. - **Visualization Support:** Plots rendered as images, displayed in the dashboard. - **Markdown Rendering:** LLM insights shown as neat, readable Markdown. --- ## Setup Instructions ### 1. Prepare Your Project Structure Upload the following folders into your project: - `templates/` (HTML templates for UI) - `static/` (CSS and static assets) - `uploads/` (for storing uploaded CSV files) ### 2. Add Your Python Files Make sure you have your main Flask app (`app.py`) and MCP server (`mcp_server.py`) in the root or `app/` folder. ### 3. Create a Python Virtual Environment ```sh python3 -m venv .venv source .venv/bin/activate ``` ### 4. Install Dependencies ```sh pip install -r requirements.txt ``` ### 5. Add Your OpenAI API Key Create a file named `.env` in the same folder as your Python files: ``` OPENAI_API_KEY=your_openai_api_key_here ``` **Never commit your `.env` file to git!** ### 6. Run the MCP Server ```sh python mcp_server.py ``` ### 7. Run the Flask App Open a new terminal, activate your environment, and run: ```sh python app.py ``` ### 8. Access the App Go to [http://localhost:5000](http://localhost:5000) in your browser. --- ## Usage Guide 1. **Upload CSV:** On the home page, upload your dataset. 2. **Analyze:** After upload, you'll see dataset columns and a question box. 3. **Ask Questions:** Type any data analysis question (e.g., "Show missing values", "Plot age vs salary", "Find outliers"). 4. **Results Dashboard:** View visualizations, summaries, and Markdown insights. 5. **Reset:** Use the reset button to upload a new dataset. --- ## Project Structure ``` AI-NEW-AGENT/ ├── app.py # Main Flask app ├── mcp_server.py # Modular MCP server ├── templates/ # HTML templates (dashboard UI) │ ├── index.html │ ├── analyze.html │ └── result.html ├── static/ # CSS and static assets │ └── style.css ├── uploads/ # Uploaded CSV files ├── .env # Your OpenAI API key (not tracked by git) └── requirements.txt # Python dependencies ``` --- ## 📸 Screenshots ### 1. Upload CSV File Easily upload your dataset to start the analysis. <img src="images/upload.png"/> --- ### 2. Ask a Question Type your query, such as *“give me any visualisation of salary”*. <img src="images/demo question.png"> --- ### 3. Visualization Example The agent generates a **Histogram of Salary** along with a clear explanation of the distribution. <img src="images/hist of salary.png"> --- ### 4. Missing Values Report Quickly check how many **null values** exist across all columns. <img src="images/null values.png"> --- ## Troubleshooting - **OpenAI API errors:** Check your `.env` and API key. - **MCP server not running:** Make sure `mcp_server.py` is started before `app.py`. - **CSV upload issues:** Ensure your file is a valid CSV and not too large. ---