Goal
Prove out real analyst SQL on a dataset realistic enough to have something genuine to say about customer behaviour.
Project
SQL Retail Analytics
A working, tested analysis with documented findings: revenue seasonality, a long-tail customer value distribution, category-leading products, and a personalised churn-risk model, all runnable and readable directly on GitHub.
Normalized relational schemaWindow functions & cohort analysisData-integrity test suite
My role
Solo build: schema design, synthetic data generation, SQL analysis, and testing
SQLPythonJupyter Notebookpytest
View source on GitHub
Result
Five standalone SQL queries answering specific business questions, each backed by an executed, documented notebook and a data-integrity pytest suite.
Support
Used a gap-ratio churn rule (a customer's current order gap versus their own historical average) instead of a flat days-since-last-order cutoff, for a more personalised at-risk list.