Employee survey

REFRUITING / DEMAND MODEL

Predict without hindsight.

Chronological train/holdout evaluation. Daily observed attendance is not fed into its own forecast.

BACKTEST MAE
—people
Rolling chronological holdout
BACKTEST WAPE
N/A%
FIXED BASELINE WAPE
N/A%
SUGGESTED VOLUME
—kg

Held-out daily attendance

→
Reading operational data…
Predicted staff; actual scored separately

Observed history / O001

No matching source rows

45-office source cutoff register

OfficeZonePredicted staffOrder / kgFeature availability

Synthetic Refruiting challenge data · simulated plan, not a live delivery service · all dimensions and values from CSV