This repository contains my completed work for the Quantium Virtual Experience Program offered on Forage in May 2025. The simulation replicates the type of work done by a data analyst in the Data Science and Analytics team at Quantium, focusing on customer behavior analysis, uplift testing, and commercial decision support.
🧠 Project Overview Quantium is a global leader in data science and AI, working across industries such as retail, banking, and health. This simulation focused on helping the commercial team make data-driven decisions using transaction-level data. The entire project was broken into 4 structured tasks, progressively building insights and culminating in a final report for the Category Manager.
✅ Tools and Skills Used Category Tools / Techniques Programming Python (Jupyter Notebook) Data Libraries pandas, numpy, matplotlib, seaborn Analytics Techniques Data Cleaning, Customer Segmentation, Uplift Testing Visualization Charts and plots using seaborn/matplotlib Reporting Markdown, Written Insights, PDF Summary Business Thinking Commercial Recommendations, Trial Store Testing, Benchmarking
🔍 Tasks Breakdown
📁 Task 1: Data Preparation and Exploration Cleaned and pre-processed transaction datasets.
Explored variables like date, store, product_id, customer_id, and sales.
Identified anomalies (e.g., irregular price patterns) and handled missing data.
Output: Clean datasets ready for analysis.
📁 Task 2: Customer Analytics
Created customer segments based on purchasing behavior.
Compared metrics across loyal vs. new customers.
Found patterns in product category preference and spending habits.
Output: Insightful visualizations and key customer behavior summaries.
📁 Task 3: Uplift Testing on Trial Stores
Identified control vs. trial stores using benchmarking logic (based on sales patterns and demographics).
Conducted uplift analysis to determine if store layout trials led to a statistically significant sales increase.
Included pre-trial matching and post-trial comparisons.
Output: Uplift test results with statistical justification.
📁 Task 4: Final Report to Category Manager
Synthesized insights from previous tasks.
Created actionable, data-backed recommendations for commercial stakeholders.
Focused on increasing customer retention, optimizing store layout, and identifying high-performing store types.
Output: Professional report for business decision-making.
❌ Not Used in the Simulation
To stay aligned with the scope of the Quantium simulation, the following were not included:
No use of machine learning models or predictive modeling
No SQL-based querying (though relevant in real-world work)
No dashboard creation (e.g., Tableau, Power BI)
No use of external datasets beyond those provided
No deployment or API integration
📌 Outcome and Learnings
This simulation helped reinforce my ability to:
Think critically about business problems using data
Clean and analyze real-world retail datasets
Evaluate experimental designs such as uplift testing
Translate analytical work into commercial recommendations
It also strengthened my technical proficiency in Python for Data Analytics, and deepened my understanding of data-driven decision-making in a retail setting