AI-Fraud-Detection-Warehouse-Monitoring

KO | EN

gitlite — search

#javascript #python #hacktoberfest #react #ai #typescript #llm #go #golang #android #machine-learning #rust #deep-learning #linux

AI-Fraud-Detection-Warehouse-Monitoring

★ 22 Open GitHub ↗

AI-powered fraud detection and warehouse monitoring system using Isolation Forest and Random Forest with explainable anomaly detection. Demo version with synthetic data.

// repository documentation

Was this content helpful?

★ 0 (0 ratings)

# AI-Based Fraud Detection & Warehouse Monitoring System
### Demo Version | Developed by Faisal Umar | [Fasai.uk](https://fasai.uk)

---

## ⚠️ Important Disclaimer
This is a **demonstration system using entirely synthetic/sample data**.
No real company data, client data, or proprietary information is used or stored.
The actual production system was deployed for a real organisation under a confidentiality agreement.

---

## 🔍 Overview

This system demonstrates an AI-powered fraud detection and warehouse monitoring solution designed to identify suspicious stock movements and irregular financial activities in warehouse operations.

**Before this system:** Warehouse monitoring was entirely manual, slow, and dependent on fragmented reporting — increasing operational risk and reducing management control.

**After deployment:** Manual intervention reduced by approximately 90%, with real-time automated fraud detection and operational monitoring.

---

## 🤖 Technical Architecture

```
Warehouse & Sales System
         |
         ▼
Data Ingestion & Preprocessing
         |
         ▼
Behavioural Feature Engineering
         |
    ┌────┴────┐
    ▼         ▼
Isolation   Random
Forest      Forest
(Anomaly)  (Classifier)
    └────┬────┘
         ▼
Combined Risk Scoring
         |
         ▼
Alert & Validation Workflow
         |
         ▼
Administrative Monitoring & Audit Logs
```

---

## 🚀 Features

- **Real-time transaction monitoring** across multiple warehouse locations
- **Dual-model detection:** Isolation Forest + Random Forest
- **Combined risk scoring** (0-100 scale) with Low/Medium/High classification
- **Fraud explanation engine** — explains WHY a transaction was flagged
- **Interactive dashboard** with filters by location, staff, and risk level
- **Feature importance analysis** — transparent and auditable
- **Full transaction investigation** tool for flagged records

---

## 📊 Fraud Detection Patterns

The system detects the following suspicious patterns:
1. **After-hours transactions** — activity outside business hours
2. **Excess stock removal** — unusually high stock outflows
3. **Price manipulation** — prices significantly below normal range
4. **Behavioural anomalies** — unusual patterns detected by Isolation Forest
5. **Statistical outliers** — transactions deviating from historical norms

---

## 🛠️ Installation & Running

```bash
# Clone the repository
git clone https://github.com/Faisal51214/noor-brands-fraud-detection-demo

# Install dependencies
pip install -r requirements.txt

# Generate sample data
python generate_data.py

# Run the dashboard
streamlit run app.py
```

---

## 📁 Project Structure

```
noor_brands_demo/
│
├── app.py                 # Main Streamlit dashboard
├── fraud_detection.py     # Core ML models and detection logic
├── generate_data.py       # Sample data generator
├── requirements.txt       # Python dependencies
├── warehouse_data.csv     # Generated sample data
└── README.md             # This file
```

---

## 🔬 Models Used

| Model | Type | Purpose |
|-------|------|---------|
| Isolation Forest | Unsupervised | Anomaly detection without labels |
| Random Forest | Supervised | Fraud probability classification |

---

## 📈 Impact Metrics (Production System)

| Metric | Before | After |
|--------|--------|-------|
| Manual intervention | 100% | ~10% (90% reduction) |
| Fraud detection speed | Hours/Days | Real-time |
| Monitoring coverage | Partial | 100% |
| Operational visibility | Limited | Full real-time |

---

## 👨‍💻 Developer

**Faisal Umar**
- 🌐 Platform: [Fasai.uk](https://fasai.uk)
- 💼 LinkedIn: [linkedin.com/in/faisalumar](https://www.linkedin.com/in/faisal-umar-43888333b/)
- 🐙 GitHub: [github.com/Faisal51214](https://github.com/Faisal51214)
- 📧 Email: faisal.umar51214@gmail.com

---

## 📄 License

This project is open source and available for educational and portfolio purposes.

AI-Fraud-Detection-Warehouse-Monitoring

Explore Similar Repositories

Was this content helpful?

Recent Feedback

Download README