calendar.csv
101 KB
1,969 date rows
The pipeline starts with the Kaggle M5 Forecasting – Accuracy
competition files. Three production source CSVs were placed under
data/raw/ as the immutable landing zone for all later Bronze work.
The competition zip contains additional files, but the data engineering project uses only the sources required for a full historical sales, calendar, and price spine.
| File | Role | Decision |
|---|---|---|
calendar.csv |
Date dimension, events, SNAP flags, Walmart week ids | Required |
sales_train_evaluation.csv |
Full daily unit sales history (wide format) | Required |
sell_prices.csv |
Weekly sell prices by store and item | Required |
sales_train_validation.csv |
Earlier subset of the same history | Skipped |
sample_submission.csv |
Forecast competition submission template | Skipped |
| Source | Columns | Notes |
|---|---|---|
| calendar | 14 |
date, wm_yr_wk, weekday, wday, month, year, d,
event_name_1/2, event_type_1/2, snap_CA/TX/WI
|
| sales | 1,947 |
Identity cols id, item_id, dept_id, cat_id, store_id, state_id
plus d_1…d_1941
|
| sell_prices | 4 | store_id, item_id, wm_yr_wk, sell_price |
walmart-m5/ ├── data/ │ └── raw/ │ ├── calendar.csv │ ├── sales_train_evaluation.csv │ └── sell_prices.csv ├── airflow/ ├── src/ ├── notebooks/ └── README.md