Research on Application of Data Mining to Sales of Goods in Chain Retail Supermarkets

Due to the rise in diversity of consumers' needs in recent years, sales of goods have become customer-oriented rather than supply-oriented. Salesmen need to be sharp and sensitive to changes in consumer behaviors in the marketplace so that they can rather quickly rearrange display of products and selections of items on promotion in order to have the edge in decision making. In this paper, data were collected starting from November 2007 which amounts to 3.4 million records. The proposed method involved integrating data mining with OLAP to find the key correlation in data dimensions and apply them to the chain super-markets. The same conditions were used to query data which were compared with results from different locations. Results showed considerable differences for different locations. For example, in Tianmu, Taipei (a tourist paradise), high correlation is shown with leisure goods. On the other hand, in Yunlin, south Taiwan (country life), reflected high correlation with daily necessities and foods. Multi-dimension market basket analysis method can analyze the sales correlation of multi-dimensions such as time, customers and products in a supermarket. It also provides salesmen with consumer behaviors or habit information in different areas. Therefore, salesmen can plan the marketing strategy and combination of items for a local area to increase sale revenues..