Forecasting Demand and Inventory

What statistical methods or analytical models do you find most effective for forecasting demand or inventory levels?

Provide a practical recommendation that distinguishes demand forecasting from inventory-policy modeling. Explain how you would select, validate, and monitor the approach when demand may contain trend, seasonality, promotions, intermittent observations, and changing variance.

Requirements

1. Compare seasonal-naive benchmarks, exponential smoothing, ARIMA or SARIMA, regression with external drivers, and intermittent-demand methods such as Croston's method.
2. Explain rolling-origin validation and appropriate error metrics.
3. Show how forecast uncertainty informs safety stock and reorder points.
4. State assumptions, limitations, and conditions requiring model recalibration.

Assumptions

Use methods appropriate to the available demand history, forecast horizon, lead-time distribution, service-level target, and item demand pattern. No numerical dataset is supplied, so express calculations symbolically.