Year-over-Year Sales Analytics
Compares sales against last year properly: discounting the stores that opened or closed in between.
The problem
Comparing against last year by hand is misleading: calendars don't align and store openings/closings distort the read.
What I did
Built an ISO comparable-time engine by day, week, month and year, with coverage-based Same-Store Sales, full dimensional filters, four analysis tabs and multi-sheet Excel export.
The outcome
In production: it is the business's standard year-over-year comparison. No generative AI — here the value is in defining what is actually comparable.
How it compares
- Period salesInput
- ISO comparable periodsDay, week, month and year
- Store coverageWhich stores existed in both
- Same-store salesThe clean comparison
- Dimensional filtersFour analysis tabs
- Multi-sheet Excel exportOutput
Comparing against last year looks trivial and is not: calendars do not line up, and store openings and closures muddy the reading. No LLM is involved here — it is pure BI, and the value lies in defining what is comparable.
Stack
- Python
- Streamlit
- pandas
- Altair
- SQL Server
- Supabase
- Docker
Blue Star Group internal project — the code is not public.
Next project: Multi-language Catalog Translator →