Three datasets, three different business questions.
This week focused on moving beyond basic exploration and using data to compare performance, profitability, customer behavior, and market structure.
The projects cover:
- Superstore profitability with SQL
- Regional video game market analysis
- UK e-commerce customer and revenue analysis
Dataset: Sample Superstore
Rows: 9,994
Period: 2014–2017
Main tools: MySQL, SQL, Jupyter
- regional and sub-category profitability
- discount impact
- shipping performance
- customer value
- year-over-year sales growth
- loss-making product groups
The biggest issue was not weak sales, but revenue converting poorly into profit in specific areas.
Examples:
- discounts above 20% were associated with much weaker profitability
- Tables generated losses in multiple regions
- Copiers produced the highest total profit
- 2016 and 2017 showed strong sales growth after a slight decline in 2015
Business question:
Where is strong sales activity failing to generate healthy profit?
Dataset: Video Game Sales
Cleaned rows: 16,327
Main tools: Python, pandas, Matplotlib, Seaborn
- genre performance
- platform rankings
- publisher rankings
- regional sales structure
- yearly sales trends
- top-selling games
- regional genre preferences
Global rankings hide important regional differences.
- Action leads globally and in North America / Europe
- Role-Playing has a much stronger relative position in Japan
- Nintendo leads publishers
- PS2 leads platforms
- 2008 is the strongest year in the dataset
Business question:
How do platform, genre, and regional preferences shape market performance?
Dataset: UK Online Retail
Raw rows: 541,909
Clean positive-sales rows: 397,924
Identified customers: 4,339
Main tools: Python, pandas, Jupyter
- customer revenue
- order frequency
- average order value
- country contribution
- monthly sales
- cancellations
- basket behavior
- product demand
The business is highly concentrated.
- the UK contributes roughly 82% of cleaned revenue
- November 2011 is the strongest revenue month
- the top 5 customers account for about 11.7% of revenue
- cancellation behavior varies significantly by country
Business question:
Which customers, markets, products, and periods drive revenue — and where does transaction behavior create risk?
Week 1 was mainly about:
inspect → clean → explore
Week 2 moved further into:
query → compare → validate → interpret
The main shift was from understanding datasets to using them to answer more specific business questions.
01-superstore-sql-sales-profit-analysis/
02-video-game-regional-sales-analysis/
03-uk-ecommerce-customer-revenue-analysis/
SQL MySQL Python pandas Matplotlib Seaborn SQLAlchemy Jupyter
Different datasets require different methods, but the analytical objective stays the same:
identify what is driving performance, test whether the pattern is real, and explain why it matters.