A sales forecasting model for new-released and nonlinear sales trend products

This paper proposes a sales forecasting model for new-released products with a knowledge-based database. Though forecasting future demand is an essential part of business planning and operation, most major forecasting methods are applied only to regular consuming items, showing gradual sales trend caused by seasonal cycles. Most retail products are, however, irregular consuming products characterized by fluctuated and nonlinear sales trends. In this study, using high correlations between short- and long-term accumulated sales within similar products groups, a new forecasting model is presented. Based on the correlation database of short- and long-term accumulations, the model provides a prediction of long-term forecast using the sales result of the product’s very early release. For practical use, the model is designed to deal with the following three points: accuracy; timing of forecast release; and the broad coverage of items. As a case study, we applied the model to books and consumer electronics sold in Japan. The model enables us to obtain a practical sales forecast throughout the lifecycle of the item one or two week after its release, and furthermore provides valuable information on reprint decision-making. This experiment will prove the reliability in accuracy and efficiency of the proposed method in comparison with existing established ones.