An intelligent, tool-augmented AI agent built with LangGraph that can analyze and visualize tabular data, such as CSV files or pandas DataFrames. It leverages LLM capabilities and Python tools to assist in data exploration, reporting, and visualization.

• 🔍 Descriptive Statistics

  • Calculate mean, median, min, max, and standard deviation.
  • Optionally group statistics by one or more columns.

• 📊 Data Visualizations

  • Supports bar plots, line plots, scatter plots, histograms, and box plots.
  • Uses matplotlib + seaborn under the hood.

• 🧠 LLM-Powered Interaction

  • Automatically decides when to compute or plot based on user instructions.
  • Intelligently asks clarifying questions if user input is incomplete.

• Computes descriptive statistics for specified columns.
• Supports optional grouping (e.g., "Sales by City").

• Visualizes relationships between numeric columns.
• Intelligently selects plotting strategy based on user input.

• "Show me the average sales by city"
• "Plot sales against footfall as a scatter plot"
• "Give me the summary stats for basket size"
• "Generate a histogram of weekly footfall"

📁 data-insights-agent/
├── agent.py              # Main LangGraph implementation
├── retail_data.csv       # Sample dataset
├── README.md
├── requirements.txt

Unlike traditional agents that rely solely on language models:

✅ Tool-augmented agents can:

• Execute actual statistical code (e.g., pandas groupby)
• Visualize data dynamically with Python plots
• Avoid hallucinating facts not in the dataset
• Ask for clarification before proceeding

This hybrid approach ensures accuracy, flexibility, and explainability—perfect for analysts, data scientists, and educators.

• Combines language understanding (LLM) with functional execution (tools)
• Can route intelligently between different tools based on user intent
• Ensures precision in logic-heavy tasks like conversions, which pure LLMs are not designed for
• Easily extensible — add more tools like date conversion, BMI calculators, tax estimators, etc.

🤖 This is not just an LLM — it's an intelligent agent that knows when to "think" and when to "act".

pip install -r requirements.txt
python agent.py

Make sure to add your CSV or use the provided retail_data.csv. The app will prompt you for natural-language queries.

Here are some sample prompts you can use to interact with the Data Insights Agent:

• "Show me the average sales by city."
• "What is the median footfall for each week?"
• "Generate a scatter plot of Sales vs. Footfall."
• "Give me summary statistics for 'Avg Basket Size'."
• "Plot a bar chart comparing average sales across different cities."
• "Show the trend of footfall over time using a line plot."
• "Which week had the highest average basket size?"
• "Visualize the distribution of footfall using a histogram."
• "Compare the sales performance grouped by week and city."

You can mix and match chart types, metrics, and grouping variables to uncover deeper insights.

Jayesh Suryawanshi

• 🧠 Python Developer | 💡 AI Tools Builder | 🌍 Data & Engineering Enthusiast
• 📫 LinkedIn

MIT License