🚀 RevStack

Retail Sales Analytics using Azure Data Lake, Databricks & Power BI

📌 Project Overview

RevStack is an end-to-end Retail Data Engineering Capstone Project designed to build a scalable analytics pipeline for sales, inventory, and customer insights.

The project leverages Azure Data Lake Gen2, Databricks (Delta Lake & PySpark), SQL, and Power BI to transform raw retail data into business-ready analytics with historical tracking and real-time alerts.

---

🎯 Problem Statement

• To design a modern data platform that enables:
  • 📈 Sales trend analysis across regions and product categories
  • 📦 Inventory monitoring and stock-out detection
  • 🕒 Historical tracking of customer and product changes (SCD Type 2)
  • 🏆 Identification of top-performing customers and products

---

🧱 Architecture Overview

Medallion Architecture (Bronze → Silver → Gold)

      Raw CSV Data (ADLS Bronze)
              ↓
      Cleaned & Typed Delta Tables (Silver)
              ↓
      Business Logic + SCD2 + Aggregations (Gold)
              ↓
      Power BI Dashboards

This architecture ensures scalability, data quality, governance, and analytics readiness.

---

🗂️ Data Sources (Bronze Layer)

• Raw CSV files stored in Azure Data Lake Gen2:
  • Customers
  • Products
  • Inventory
  • Sales
  • Regions
• 🔗 Dataset Link: Dataset

---

🧹 Data Cleansing – Silver Layer

• Transformations Performed:
  • Schema enforcement (data type casting)
  • Null handling & duplicate removal
  • Date formatting
  • Conversion to Delta Tables
• Silver Tables:
  • dim_customer
  • dim_product
  • dim_inventory
  • dim_region
  • fact_sales

---

🏗️ Data Modeling – Gold Layer

• Gold Tables Created:
  • dim_customer_scd2 – Customer historical tracking
  • dim_product_scd2 – Product change history
  • fact_sales – Transactional sales data
  • inventory_alerts – Low-stock monitoring

---

🔁 Slowly Changing Dimension (SCD Type 2)

• Tracks historical changes in customer and product attributes
• Maintains:
  • start_date
  • end_date
  • is_current
• Supports Delta Time Travel for versioned data analysis

---

🚨 Inventory Alerts Logic

• Business Rule: - IF StockQuantity < MinThreshold → Raise Alert
  • Alerts stored as Delta table in Gold layer
  • Enables near real-time stock-out visibility

---

📊 Power BI Dashboard

• Key Dashboards & Insights:

  📈 Sales Trends

  • Monthly sales by Region & Category
  • Filters: Product, Region, Quarter

  📦 Inventory Levels

  • StockQuantity vs Product
  • Highlighted low-stock alerts

  🕒 Historical Customers (SCD2)

  • Active & historical customer records
  • Change tracking over time

  🏆 Top Performers

  • Top customers by lifetime sales
  • Best-selling products

  🔁 All visuals are fully interactive and filter-driven

---

👉 Power BI Report (View Only):

For a detailed report and interactive analysis, refer to the RevStack

---

🛠️ Tech Stack

Component	Technology
Storage	Azure Data Lake Gen2
Processing	Azure Databricks
Framework	PySpark, Delta Lake
Data Modeling	Medallion Architecture
Governance	Unity Catalog
Analytics	SQL
Visualization	Power BI

---

✅ Key Outcomes

• ✔ End-to-end retail data pipeline implemented
• ✔ SCD Type 2 historical tracking enabled
• ✔ Real-time inventory alerting logic
• ✔ Curated Gold layer for BI reporting
• ✔ Enterprise-grade Power BI dashboard

---

🧠 Conclusion

RevStack demonstrates a real-world data engineering architecture for retail analytics, combining scalable pipelines, historical tracking, and actionable business insights — ready for future enhancements like ML forecasting or automated alerts.