Rather than automatically proceeding to forecast with data at the same level of aggregation as that required for an organizationÕs operations, the authors explain that the best level of aggregation for forecasting should be chosen by the forecasters in consideration of the trade- off between sampling error (data inadequate to generate reliable forecasts) and specification error (data too aggregated to represent diverse demands).
Forecasting by cross-sectional aggregation / Zotteri, Giulio; Kalchschmidt, Matteo; Saccani, Nicola. - In: Foresight: The International Journal of Applied Forecasting. - 35(2014), pp. 35-41.