Weighted MAPE
Compute WMAPE = SUM(ABS(error)) / SUM(actual) for STATISTICAL. Single number.
| Column | Type | Key |
|---|---|---|
| actual_id | INT | PK |
| sku_id | INT | FK → dim_skus |
| location_id | INT | FK → dim_locations |
| period_date | TEXT | |
| units_sold | INT | |
| revenue | REAL |
| Column | Type | Key |
|---|---|---|
| forecast_id | INT | PK |
| sku_id | INT | FK → dim_skus |
| location_id | INT | FK → dim_locations |
| version_id | INT | FK → dim_forecast_versions |
| forecast_period_date | TEXT | |
| forecast_generated_date | TEXT | |
| forecast_qty | INT |
Each hint you reveal reduces the XP you can earn. Try the query first.
WMAPE weights errors by volume — the proper aggregation is sum-of-absolute-errors over sum-of-actuals (not the average of MAPEs).
SUM(actual) denominator.
Force float.