Skip to content

End-to-end Business Intelligence solution using Snowflake, SQL, and Tableau to identify performance gaps and recommend high-ROI market expansion strategies based on 180K+ transaction records.

Notifications You must be signed in to change notification settings

apekshatej/Business-Intelligence-solution-for-Retail

Repository files navigation

Business Intelligence Data Warehouse Project

Overview

This project was developed as part of IMT 577: Business Intelligence Systems.
It focuses on building a dimensional data warehouse to support analytical insights for retail store operations, addressing:

  • Sales performance analysis
  • Fair bonus allocation
  • Weekday sales trends
  • Market expansion recommendations

The solution integrates raw transactional data, sales targets, and master data into a dimensional model using Snowflake.
Secure views were created for access, and Tableau dashboards were built for interactive analysis.


Objectives

  • Assess store-level performance against sales targets.
  • Recommend fair allocation of a $2M bonus pool.
  • Analyse weekday sales patterns for efficiency.
  • Evaluate expansion opportunities.
  • Deliver dashboards for data-driven decision-making.

Project Schedule

  • Apr 23, 2025 – Project planning & scope
  • May 1, 2025 – Data profiling & source identification
  • May 7, 2025 – Dimensional model design (Star Schema, ERD, Fact & Dimension tables)
  • May 14, 2025 – Data transformation & loading
  • May 21, 2025 – Data access layer & secure views
  • May 27, 2025 – Dashboard development & insights
  • May 31, 2025 – Group synthesis & presentation prep
  • Jun 6, 2025 – Final presentation & recommendations

Data Architecture

Fact Tables

  • Fact_SalesActual – transaction-level sales
  • Fact_ProductSalesTarget – product-level sales targets
  • Fact_SRCSalesTarget – channel/store-level targets

Dimension Tables

  • Dim_Date – time attributes
  • Dim_Product – product details
  • Dim_Store – store details
  • Dim_Customer – customer profiles
  • Dim_Reseller – reseller details
  • Dim_Channel – sales channel classification
  • Dim_Location – location and geography

Design Notes

  • Snowflake schema with normalized location data
  • Surrogate keys for performance & integrity
  • Secure views for reporting & Tableau integration

Analysis Framework

Business questions addressed

  1. How are Stores 10 & 21 performing vs. targets?
  2. How should a $2M bonus pool be allocated fairly?
  3. What weekday sales patterns can be identified?
  4. Should new stores be opened, and where?

Tools & Technologies

  • Azure Blob Storage – source data storage
  • Snowflake – data warehouse (ELT, schema, secure views)
  • SQL – transformations, data cleansing, fact/dim table creation
  • Tableau – dashboards for analysis & insights

Dashboards

Built with Tableau for dynamic insights:

  • Store performance vs. targets
  • Bonus allocation distribution
  • Weekday sales patterns
  • Expansion opportunities

Dashboard Link: Store Performance


Key Insights

  • Stores 10 & 21 underperformed in 2014, achieving ~82% and ~76% of targets
  • Both showed positive growth trends, so closures were not recommended
  • Bonus allocation was proportional to performance ratios, rewarding high achievers
  • Weekday sales peaked on Mondays & Sundays, dipped midweek
  • Expansion analysis identified Georgia (Store 5) as a high-potential location

How to Run

  1. Clone or download project files.
  2. Load raw CSV data into Snowflake staging tables (via Azure Blob Storage).
  3. Run SQL scripts to:
    • Create Dimension & Fact tables.
    • Load transformed data.
    • Create secure views for reporting.
  4. Connect Tableau to Snowflake views for dashboarding.

License

This project was developed for academic purposes under IMT 577: Business Intelligence Systems.
It is not licensed for commercial use.


About

End-to-end Business Intelligence solution using Snowflake, SQL, and Tableau to identify performance gaps and recommend high-ROI market expansion strategies based on 180K+ transaction records.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published