Superstore Sales Dataset Csv Download Data: Dataset Structure and Field Coverage

Eighteen columns cover the full order-line story, and Datadory verified each field definition from the CSV header and observed values rather than documentation:

Office Supplies dominates the file: 5,909 of 9,800 rows, or about 60 percent, against Furniture at 2,078 and Technology at 1,813. Regionally the lines spread West (3,140), East (2,785), Central (2,277) and South (1,598) across United States shipping addresses, so a four-region by three-category cross-tab - twelve cells, all populated - falls straight out of the header row.

Coverage runs from 3 January 2015 to 30 December 2018: exactly four calendar years of daily-dated transactions, static since publication. That window is long enough for monthly seasonality decomposition and annual comparisons, and short enough that every model trained on it inherits a pre-2019 retail world.

The category's weight inside the file is itself the reason forecasters keep it: Office Supplies lines are small-ticket, high-frequency purchases, which produces a steadier monthly series than furniture's lumpy big-ticket orders. If your exercise needs noise, model Furniture; if it needs a stable seasonal signal, the 5,909-row subset delivers it.

The file loads fast enough that a first forecast lands inside an hour:

That last point is the honest boundary: the dataset remains the classic forecasting benchmark because it is small, clean and fully labeled, not because it predicts anything current.

Three complements extend the file without changing its schema:

Joins are fuzzy by nature. Match Office Depot SKUs to Superstore products on brand and name tokens and expect partial overlap, since one side is a fictionalized 2015-2018 sample and the other is a live chain's current shelf. For the market series, align on month and note that NAICS category definitions will not map one-to-one onto the file's own three categories.