Short-Term Sales Forecast of Perishable Goods for Franchise Business

In a franchise chain system, the franchise organization grants the right to the franchisee to represent and sell its products, but franchisee can only order products from them. Due to the buy-out policy of ordering, franchisee store managers perform demand forecasting and inventory control based on their sales performance and experience. While the products are perishable goods with a very-short preservation time, the ordering strategy tends to be more conservative, to avoid the unsold items that will turn into wastes. This conservative strategy not only limits the profit but also makes the business end earlier. A case study of a bakery franchise business in China which sell bread with one-day preservation time was investigated in this research. More than 10 million of point-of-sales (POS) data were collected and analyzed. In order to provide a better demand forecasting in a relatively short period, we utilize the number of sales in the first few hours and some supporting variables as explanatory variables to predict the number of sales for the rest hours in a single operation day. The regression analysis was compared with one of machine learning method: neural network techniques as the forecasting tools. The experimental result shows that the number of sales in the first few hours can be used to predict the rest of sales in a single day. This approach provides a systematic forecasting model to reduce the waste of the supply chain system.