📊 Quantium Retail Analytics Project

📌 Overview

This project is part of the Quantium Retail Strategy & Analytics Virtual Internship, where the goal was to analyze retail transaction data, identify customer insights, test trial store strategies, and provide commercial recommendations for the chips category.

The project is divided into three key tasks:

1. Task 1 – Data Preparation & Customer Analytics
2. Task 2 – Experimentation & Uplift Testing
3. Task 3 – Analytics & Commercial Application

The final deliverable is a client-ready report built using the Pyramid Principle framework, including actionable recommendations for the Category Manager.

---

📝 Task Details

🔹 Task 1 – Data Preparation & Customer Analytics

• Merged transaction and customer data.
• Cleaned and formatted dataset (dates, product names, outliers).
• Segmented customers by lifestage and premium status.
• Key Insights:
  • Young Singles/Couples and Older Families are the main buyers.
  • Customers prefer large pack sizes (175g+).
  • Premium customers spend more on average.

---

🔹 Task 2 – Experimentation & Uplift Testing

• Conducted trial vs control store analysis for stores 77, 86, and 88.
• Metrics compared:
  • Total sales revenue
  • Number of customers
  • Average transactions per customer
• Used Pearson correlation & magnitude distance to select control stores.
• Key Findings:
  • Store 77 → Significant uplift in sales.
  • Store 86 → Moderate uplift.
  • Store 88 → No significant uplift.
• Uplift was mainly driven by more customers, not more transactions per customer.

---

🔹 Task 3 – Analytics & Commercial Application

• Built a client-facing report using the Pyramid Principle framework.
• Presented insights with charts, callouts, and recommendations.
• Final Recommendations:
  • Scale promotions from Stores 77 & 86 to similar stores.
  • Focus on Young Singles/Couples segment.
  • Invest in large pack promotions.
  • Reassess Store 88 strategy before expanding further.

---

📊 Key Visuals

• 📈 Customer segmentation charts
• 📊 Trial vs Control store sales comparisons
• 📉 Distribution plots of sales & customers

---

⚙️ Tech Stack

• Python: pandas, numpy, matplotlib, seaborn
• Jupyter Notebook for analysis
• PowerPoint / PDF for final report
• GitHub for version control

---

📌 Learning Outcomes

• Hands-on experience in data cleaning, analytics, hypothesis testing.
• Ability to design uplift experiments and evaluate trial store performance.
• Developed storytelling skills using data → client-ready presentation.

---

👤 Author

Tarun Kumar Malviya
📧 tarunmalviya804@gmail.com