📊 Auto EDA Agent

Auto EDA Agent is an Agentic AI-powered Exploratory Data Analysis (EDA) system that performs end-to-end data analysis on CSV datasets.
It combines deterministic Python analytics with LLM-driven reasoning and insight generation, following a true plan → act → observe → explain agentic workflow.

This project is designed to demonstrate production-grade Agentic AI architecture, not just automated plotting.

---

🚀 Key Features

• 📁 Upload any CSV dataset
• 🎯 Define analysis goals in natural language
• 🧠 LLM-powered EDA planning (no hardcoded steps)
• 🔧 Deterministic tool execution layer (Pandas, Matplotlib, Seaborn)
• 📊 Curated, high-signal visualizations (not chart spam)
• 📉 Intelligent chart prioritization (max 5–8 charts)
• 🔍 Grounded insights with actual numeric facts
• 🧩 Clean separation of reasoning, execution, and explanation
• 🖥️ Interactive UI built with Streamlit

---

🧠 Agentic AI Architecture

This project follows a multi-phase Agentic AI design, inspired by real-world AI systems.

🔹 Phase 1 — UI Layer

• Streamlit-based interface
• Dataset upload, goal input, and results display

🔹 Phase 2 — Data Intake

• CSV ingestion using Pandas
• Schema detection and preview

🔹 Phase 3 — Reasoning (Planning Agent)

• Gemini LLM generates a step-by-step EDA plan
• No code execution
• No insights generated at this stage

🔹 Phase 4 — Execution (Tool Agent)

• Plan steps mapped to deterministic Python tools
• Controlled execution with idempotency guards
• Charts generated only once per analysis
• Visualization limits enforced

🔹 Phase 5 — Observation & Insight Agent

• Python computes factual metrics (missing %, distributions, survival rates, etc.)
• LLM generates grounded insights
• Strict prompt rules prevent hallucination
• Every insight references numeric facts

---

🛠️ Tech Stack

• Python
• Streamlit — UI & interaction
• Pandas — Data processing
• Matplotlib / Seaborn — Visualizations
• Google Gemini (LLM) — Planning & insight generation

📜 License

This project is licensed under the MIT License.

You are free to use, modify, and distribute this project with proper attribution.

---

📂 Project Structure

.
├── app.py                 # Main Streamlit application (Agentic AI pipeline)
├── requirements.txt       # Python dependencies
├── README.md              # Project documentation
├── LICENSE                # MIT License
├── .gitignore             # Git ignore rules
└── .streamlit/
    └── secrets.toml       # API keys (ignored by git)