5 Forecasting for Operations Free Online FlashCards

What operational decisions do demand forecasts support?

Demand forecasts help managers plan purchasing, production, staffing, inventory, transportation, and capacity.

What is a demand forecast—and what is it not?

A forecast is an estimate of demand, not a sales target or a guarantee.

What should be defined before choosing a forecasting method?

Specify the item or service, time period, forecast horizon, and required level of detail, such as weekly demand for one product at one warehouse.

When are judgmental forecasting methods especially useful?

Judgmental methods use expert estimates, sales-force input, customer surveys, or structured approaches such as Delphi. They are useful for new products or major changes with little relevant history.

How does a seasonal naïve forecast differ from a naïve forecast?

A naïve forecast uses the latest observation for the next period; a seasonal naïve forecast uses demand from the matching period in the previous seasonal cycle.

How does exponential smoothing weight recent observations?

Moving averages use recent observations; exponential smoothing gives newer observations greater weight. Smoothing methods can also represent trend and seasonality.

What do causal forecasting models relate demand to?

Trend, regression, and causal models relate demand to time or drivers such as price, promotions, weather, or economic conditions. The drivers must be relevant, and their future values must be available or estimable.

When should judgment be added to a statistical forecast?

Combine a statistical forecast with judgment only when specific, relevant information is missing from the model, such as a confirmed promotion.

How is forecast error defined here, and what does a positive error mean?

Using this convention, forecast error is actual demand minus forecast: $e_t = \text{actual demand}_t - \text{forecast}_t$. A positive error means demand exceeded the forecast.

What does mean absolute error (MAE) measure?

Mean absolute error is the average absolute forecast error: $\text{MAE} = \text{average}(|e_t|)$. It reports typical error in demand units.

What is the effect of squaring errors in MSE and RMSE?

MSE is the average squared error; RMSE is its square root. Squaring gives large misses greater influence, while RMSE is expressed in the original demand units.

When can MAPE be undefined or misleading?

MAPE averages absolute errors relative to actual demand: $\text{average}(|e_t / \text{actual demand}_t| \times 100\%)$. It is undefined when actual demand is zero and can mislead when demand is near zero.