high-fidelity-retail-pos-dataset-2M

A high-fidelity synthetic retail POS (Point of Sale) transaction dataset containing over 2,000,000+ rows of highly realistic, structurally sound market data. Generated using a custom Prolog simulation engine, this dataset is designed for database stress-testing, query optimization, and retail machine learning models.

🌟 Key Features

• Massive Scale: 2,000,000+ synchronized records across multiple relational schemas.
• High-Fidelity Logic: Zero random noise. Timestamps, quantities, subtotal calculations, member discounts, and AI-driven discounts follow logical business rules.
• Multi-Format Export: Available in production-ready formats: .sql (Transactional Dump), .json (NoSQL/API Mocking), .csv (Data Science/Pandas), and .pl (Prolog Fact Base).

📊 Database Schema (International Standard)

The dataset is exported using a clean, normalized global schema:

1. transactions (Main Header)

Stores the payment and summary details of each checkout.

• tx_id: Unique transaction identifier (String/Key)
• subtotal: Gross amount before discounts
• total_paid: Final amount paid by the customer
• vat_amount: Tax component (PPN equivalent)
• member_discount / ai_discount: Promotional cuts applied
• payment_method: Payment type (Cash, Digital Wallet, etc.)

2. transaction_details (Line Items)

Stores the specific products bought inside each transaction.

• tx_id: Relates to the main transaction
• barcode: Product barcode identifier
• qty: Quantity purchased
• sales_price: Unit price of the item

---

🚀 Get the Full 2,000,000 Rows Dataset

This repository contains a lightweight sample for verification.

To download the full, production-scale database containing 2 Million rows in all 4 formats (+ Lifetime schema updates), purchase a developer license here:

👉 DOWNLOAD FREE 1M DATASET ON GUMROAD

For custom volume generation or enterprise inquiries, feel free to contact the author.