Hi everyone, I’m sharing a new dataset covering 408,000 holiday purchases ($79.4M total sales) across the 2023–2025 holiday seasons.
It's 100% synthetic. I generated it with GenerateRealData, a tool I'm building, so no real shoppers or companies are included.
I built it so the numbers add up: 23 rules are checked with SQL and all of them hold. For example, unit_price equals list price minus the discount, line totals add up, and in-store purchases have no shipping time. It's a single 30-column table (86 MB CSV / 102 MB SQLite).
A few patterns baked in:
Peak hours vary: 7 am on Black Friday, 7 pm on Cyber Monday, and 8 pm on Thanksgiving.
Channel shifts: 35% of Black Friday purchases are in-store, compared to just 3% on Thanksgiving.
Discount-driven returns: Return rates jump from 12% (full price) to 41% (when discounted 50%+).
Good for: SQL practice (window functions, YoY pivots), holiday BI dashboards, and training ML return models.…