The most commonly known demand forecasting tools in practice involve smoothing techniques – e.g., simple moving average (SMA) and single exponential smoothing (SES). These assume that demand generally occurs in every period. When there are time intervals with no demand occurrences for an item of inventory, demand is said to be
. Demand is said to be
when it is intermittent and there are large variations in the sizes of actual demand occurrences. Intermittent or lumpy demand has been observed in both manufacturing and service environments. Croston’s method, developed for intermittent demand forecasting, was later found to have a positive bias due to an error in the mathematical derivation of expected demand. Approximate and ‘exact’ correction factors have been proposed. We report on an ongoing empirical investigation of statistical forecast accuracy and inventory control performance when applying SMA, SES, Croston’s method and its approximate and exact corrections.