Disaggregating Forecasts: Fairchiid Semiconductor’s Experience

Disaggregating Forecasts: FairchilD semiconDuctor’s experience By Joseph ogrodowczyk Forecasts prepared at the highest level of aggregation are generally the most accurate … the disaggregation method that accounts for seasonality further improves the disaggregated forecasts … describes the method for disaggregating a forecast that accounts for seasonality. I I n preparing forecasts, the forecaster must decide at what levels (SKU, category, or aggregate) forecasts are needed and which approach to use (top-down, middle-out, or bottom-up) in preparing forecasts. In the top-down approach, forecasts are prepared first at the highest level of aggregation (company as a whole) and then disaggregated into categories (families) and SKUs. In the bottom-up approach, forecasts are prepared first at a SKU level, and then aggregated to generate category-level and aggregate-level forecasts. In the middle- out approach, forecasts at a category level are generated first. Then, to develop a forecast at a company level, forecasts of all the categories are added. To generate forecasts at a SKU level, the forecast of each category is disaggregated. motiVations BehinD the topDown approach A firm ...