FORECASTING A CUMULATIVE VARIABLE USING ITS PARTIALLY ACCUMULATED DATA

Several forecasting algorithms have been proposed to forecast a cumula tive variable using its partially accumulated data. Some particular ca ses of this problem are known in the literature as the ''style goods i nventory problem'' or as ''forecasting shipments using firm orders-to- date'', among other names. Here we summarize some of the most popular techniques and propose a statistical approach to discriminate among th em in an objective (data-based) way. Our basic idea is to use statisti cal models to produce minimum mean square error forecasts and let the data lead us to select an appropriate model to represent their behavio r. We apply our proposal to some published data showing total accumula ted values with constant level and then to two actual sets of data per taining to the Mexican economy, showing a nonconstant level. The forec asting performance of the statistical models was evaluated by comparin g their results against those obtained with algorithmic solutions. In general the models produced better forecasts for all lead times, as in dicated by the most common measures of forecasting accuracy and precis ion.