Application of tramo-seats automatic procedure for forecasting sporadic and irregular demand patterns with seasonality

Managing sporadic and irregular demand patterns represents a relevant issue in several industrial contexts. Two main aspects have to be underlined due to their prominence: the former is the problem of forecasting future demand profiles, and the latter choosing and determining the best re-order policy to be applied, in accordance with information gained during the forecasting step. In this paper the former issue is discussed, by focusing on the management of items with sporadic and irregular demand patterns that also present a seasonality component. TRAMO-SEATS is a versatile procedure that allows quick identification of the best SARIMA forecasting model from an available set. Results obtained by its implementation are compared with those obtained by the Croston (1972) and Syntetos-Boylan (2005) methods, which represent two modified versions of simple exponential smoothing, introduced in literature for forecasting mean demand size per period specifically in case of irregular and sporadic demand profiles. In particular, two items are analysed, with the aim of demonstrating that when the strict hypothesis required by Croston’s and Syntetos- Boylan’s approaches fails, alternative forecasting methods could be required. TRAMO-SEATS represents a promising and user-friendly option.