README.md

metadata

title: UIDAI Project S.A.T.A.R.K
emoji: 🚀
colorFrom: red
colorTo: red
sdk: docker
app_port: 8501
tags:
  - streamlit
pinned: false
short_description: Data-Driven Innovation for Aadhaar

🛡️ Project S.A.T.A.R.K: AI-Powered Fraud Detection for UIDAI

Context-Aware Anomaly Detection System for Aadhaar Enrolment Centers > Team ID: UIDAI_4571 | Theme: Data-Driven Innovation for Aadhaar

---

🎯 Quick Links

• 📊 Live Analysis Notebook: Open in Google Colab
• 🚀 Live Dashboard: Hugging Face Spaces
• 📖 Project Report: View PDF
• 💻 Source Code: Available in this repository

---

🎯 Overview

Project S.A.T.A.R.K (Statistical Anomaly Tracking & Aadhaar Risk Kit) is a revolutionary fraud detection system designed to solve the critical "Accuracy vs. Fairness" trade-off in Aadhaar vigilance.

The Problem

India's demographic diversity makes global rules ineffective:

• ❌ Strict Rules: Flag legitimate activities in tribal belts (False Positives).
• ❌ Lenient Rules: Miss sophisticated fraud in metropolitan areas (False Negatives).

Our Innovation: District Normalization

Instead of using a national average, S.A.T.A.R.K compares each enrolment center against its local district baseline.

• Example: In a tribal district where late enrolment is common (Avg: 40%), a center doing 90% is flagged. But in a city where 90% is normal, it is marked safe.

---

✨ Key Features

🧠 The "Context-Aware" AI Engine

• Algorithm: Isolation Forest (Unsupervised Learning)
• Smart Logic: Detects anomalies relative to local geography.
• Capabilities: Identifies "Ghost IDs", "Sunday Surges" (Illegal Camps), and "Mass Update Operations".

📊 The Vigilance Dashboard

• Geospatial Intelligence: Interactive Heatmap of High-Risk Centers.
• Actionable Insights: "Priority Action List" exportable for field agents.
• Evidence-Based: Charts proving why a center was flagged (e.g., Weekend Activity vs. Weekday).

📥 Smart Data Ingestion

• Automated: Recursively fetches and merges fragmented CSV chunks.
• Robust: Handles massive datasets without data loss using Outer Joins.

---

🚀 Quick Start

Option 1: Run Analysis (Google Colab)

To see the Feature Engineering and Model Training in action:

1. Open the Notebook.
2. Run all cells to process the raw data.
3. Download the generated analyzed_aadhaar_data.csv.

Option 2: Run Dashboard (Local)

Prerequisites: Python 3.8+, pip

1. Clone the repository

   git clone [https://huggingface.co/spaces/lovnishverma/UIDAI](https://huggingface.co/spaces/lovnishverma/UIDAI)
   cd UIDAI
   

2. **Install dependencies**
```bash
pip install -r requirements.txt

3. Launch the App

streamlit run app.py

4. Access the Dashboard Open http://localhost:8501 in your browser.

---

📁 Project Structure

UIDAI/
├── README.md                                 # This documentation
├── requirements.txt                          # Python dependencies
├── Dockerfile                                # Container configuration
├── app.py                                    # Streamlit Dashboard Code
├── UIDAI_4571_(PROJECT_S_A_T_A_R_K_AI).ipynb # Main Analysis Notebook
├── analyzed_aadhaar_data.csv                 # Processed Data for Dashboard
├── Final-Project-Report.pdf                  # Complete Project Documentation
└── assets/                                   # Images and logos

---

🧠 Technical Architecture

The Pipeline

1. Ingestion: SmartLoader class merges fragmented CSVs.
2. Context Engine: Calculates ratio_deviation (Center vs. District).
3. AI Model: IsolationForest detects statistical outliers.
4. Visualization: Streamlit app renders the RISK_SCORE on maps.

Core Risk Signals

Feature	Logic	Detects
Ratio Deviation	(Center_Ratio - District_Avg)	Ghost IDs
Weekend Spike	Activity on Sunday / Normal Day	Illegal Camps
Mismatch Score	`	Bio - Demo
Volume Anomaly	Total_Activity > 99th Percentile	Mass Operations

---

📊 Dashboard Preview

1. Geographic Heatmap

Instantly spot high-risk clusters across India. (See assets/ for screenshots)

2. Priority Action List

Downloadable CSV for vigilance officers containing only the top 1% critical cases.

3. AI Insights Panel

"Why is this flagged?" - The AI explains its decision (e.g., "Flagged due to 500% spike in weekend activity").

---

👥 Team UIDAI_4571

Team Leader: Aman Choudhary (NIELIT Ropar)

Team Member: Prateek Dhar Dwivedi (NIELIT Ropar)

Mentor: Lovnish Verma (Project Engineer, NIELIT Ropar)

Competition: UIDAI Hackathon 2026

Submission Date: January 2026

---

📝 License

This project is open-source under the MIT License.

---

Project S.A.T.A.R.K.

Statistical Anomaly Tracking & Aadhaar Risk Kit

Built with ❤️ for a safer, inclusive Digital India.