Example with fictional data
Forecast
Demand by SKU and DC, with low, base and stress scenarios.
forecast
Pomarrosa Ion 1 L (Tetra) at CEDIS Monterrey: in week of Nov 16, with El Buen Fin 2026, the P90 could rise 4% over the weeks without an event, to 55,759 L.
Base: the non-event weeks preceding it in the horizon.
Pronóstico Ion 1 L en CEDIS Monterrey
Litros por semana y evento
Loading data
Ejemplo con datos ficticios
Ion 1 L en CEDIS Monterrey, día por día del mes
Litros por día
Loading data
Ejemplo con datos ficticios
Demanda nacional por semana
Litros por semana
Loading data
Ejemplo con datos ficticios
Demanda por SKU y zona en el horizonte
Litros por sku y zona
Loading data
Ejemplo con datos ficticios
Accuracy vs baseline
Evaluated with synthetic data4.8% less error than the baseline
Model: timestral. Baseline: the same week last year, adjusted for trend.
P10 to P90 coverage: 92.6%, against 80.0% expected
| Horizon | Skill | Weighted error (WQL) | Scaled error (MASE) | P10 to P90 coverage |
|---|---|---|---|---|
| All | 4.8% | 0.078 | 0.81 | 92.6% |
| Weeks 1 to 4 | -65.7% | 0.105 | 1.19 | 79.6% |
| Weeks 5 to 8 | 2.7% | 0.065 | 0.62 | 99.3% |
| Weeks 9 to 13 | 40.0% | 0.067 | 0.67 | 97.6% |
Measured over the last 13 closed weeks, which the model did not see when forecasting.
The demo sales are fictional: this measures the model on this example, not on your operation.