Back to Portfolio

Walmart Black Friday Sales Analysis

Company Walmart
Role Data Analyst / Data Scientist
Timeline Jul 2024 – Aug 2024
Domain
Business Analytics Statistical Analysis Retail
Skills
Hypothesis Testing Confidence Intervals Customer Segmentation EDA

Statistical deep-dive into Walmart Black Friday purchase behavior — specifically whether spending habits differ between male and female customers. Used hypothesis testing, confidence intervals, and Central Limit Theorem simulations, then packaged the findings into an interactive Streamlit dashboard with business recommendations for marketing and inventory planning.

Statistical Inference — hypothesis testing, confidence intervals, CLT simulations
Exploratory Data Analysis — spend patterns by gender, age, city, and occupation
Customer Segmentation — high-value demographic and geographic cohorts
Dashboarding — Streamlit app with visuals and stakeholder-ready recommendations

1. Overview

The core question was simple: do male and female customers spend differently on Black Friday at Walmart — and if so, by how much, and with what confidence?

Around that question I built a full statistical workflow: clean the transactional data, explore spend across demographics, run formal tests, quantify uncertainty with confidence intervals, and show how sample size affects reliability via CLT simulations. The output wasn’t just a notebook — it was a Streamlit dashboard stakeholders could actually use.

Goal: turn rigorous stats into decisions — who to target, where spend concentrates, and how confident we can be before betting marketing budget on a segment.

2. Key analysis areas

3. Visuals from the analysis

A few charts from the exploratory and statistical work. Click any image to open full size.

4. What I did

  1. Collected and preprocessed Walmart Black Friday transactional data for analysis-ready quality
  2. Ran EDA in Python to surface spend trends across customer segments
  3. Hypothesis-tested gender differences in spending for statistical significance
  4. Built confidence intervals for spending estimates at multiple confidence levels
  5. Simulated the Central Limit Theorem to show how sample size affects precision
  6. Segmented customers by age, occupation, city category, and marital status
  7. Shipped an interactive Streamlit dashboard with visuals and stats tools
  8. Wrote data-driven recommendations for marketing, inventory, and revenue
  9. Translated technical findings into language non-technical stakeholders could act on

5. Key findings

  • Spending patterns differed significantly between male and female customers
  • CLT simulations made the sample-size ↔ reliability tradeoff concrete for stakeholders
  • Age-group analysis pointed to the demographic bands with the highest spend potential
  • Geographic splits showed clear regional variation in Black Friday behavior
  • Occupation-based cuts suggested practical levers for targeted marketing

6. Tech stack

Statistical analysis

Hypothesis Testing Confidence Intervals Central Limit Theorem T-Test Descriptive Stats Sample Size Analysis

Data analysis

Data Preprocessing EDA Customer Segmentation Gender / Age / Geo Analysis Occupation Analysis

Business intelligence

Interactive Dashboards Data Visualization Business Recommendations

Programming & tools

Python Streamlit Pandas NumPy Plotly SciPy Matplotlib Seaborn