README.md

metadata

title: EDA Explorer
emoji: 📊
colorFrom: blue
colorTo: green
sdk: gradio
sdk_version: 5.16.0
python_version: '3.10'
app_file: app.py
pinned: false

🚀 EDA Explorer – AI-Powered Data Analysis CLI

A lightweight CLI tool that automates exploratory data analysis (EDA) with intelligent insights, feature importance detection, and data quality checks.

Designed to simulate how an AI Data Analyst works on real-world datasets used in EDA.

---

⚡ Key Highlights

• 🔍 One-command analysis → analyze <dataset>
• 🧠 Auto target detection for ML-based insights
• 📈 Feature importance (no manual setup)
• ⚠️ Smart data warnings (missing, ID columns, constants)
• 📊 Correlation & outlier detection
• 📁 Auto report generation (.txt)
• ⚡ Efficient handling of large datasets (Parquet + sampling)

---

🎬 Demo

👉 Full demo: https://github.com/user-attachments/assets/7dff8329-71e8-4bca-ad01-404e75df8314

https://github.com/user-attachments/assets/7dff8329-71e8-4bca-ad01-404e75df8314

---

📊 Example Output

Top Correlations:

• age ↔ income: 0.72
• tenure ↔ balance: 0.65

⚠️ Data Warnings:

• customer_id → likely ID column
• income → 52% missing values

📈 Feature Importance:

• age: 0.41 (strong signal)
• tenure: 0.32 (strong signal)

---

🧠 What Makes It Stand Out

• Automatically identifies useful vs irrelevant features
• No manual preprocessing required
• Mimics real-world data analyst reasoning
• Built using a modular agent-based system

---

⚡ Performance

• Parquet-based storage for faster I/O
• Sampling strategy for large datasets

---

🛠️ System Design

• Command handler
• Dataset registry
• Modular agents (AnalysisAgent, etc.)
• Logger integration

---

📦 Datasets

• Titanic
• Customer Churn
• Credit Card Fraud

---

🛠️ Tech Stack

• Python
• Pandas, NumPy
• Scikit-learn
• Parquet

---

🚀 Future Enhancements

• RAG-based EDA advisor
• SQL query assistant
• Model training pipeline