ORIGINAL COMPANY-CONTEXT PRACTICE
Forecast rural-service demand around a changing calendar
Original practice scenario: A hypothetical branch network plans staffing for agricultural banking services with seasonal peaks and moving holidays. Design a baseline and backtest without assuming the same date pattern each year.
Written for independent practice using company context; this is not a verified question from a company interview.
A REASONED APPROACH
Model answer.
Align forecasting with service-volume decisions and known calendar information. Agricultural seasons, holidays and branch closures can affect observed demand differently, so one repeated monthly average may be insufficient.
Work through it
- Define branch-day or branch-week demand, separating closures, missing reports and genuine zero activity; retain calendar features known before the planning deadline.
- Compare seasonal-naive and simple calendar-aware baselines through rolling-origin backtests at the staffing horizon, preserving entire future periods as holdouts.
- Report errors by branch size and peak periods, with uncertainty and separate coverage for branches lacking comparable history.
Tradeoffs to discuss
- Pooling branches stabilizes sparse series but may obscure local seasonality; individual models need enough history and monitoring.
Common pitfalls
- Do not use realized harvest or later weather information when testing an earlier planning decision.
- Observed visits may understate demand when limited opening hours or capacity prevented service.