Evaluation of the Shopping Path to Distinguish Customers Using a RFID Dataset

In this study, the authors use radio-frequency identification (RFID) data, which show the position of a shopping cart through an RFID tag attached to the shopping cart. The RFID data contain valuable information for marketing, such as shopping time and distance as well as the number of shelf visits. The authors analyze customers’ purchasing behavior and in-store movement information using POS data combined with RFID data. The purpose of this study is to discover a promising shopping path that can distinguish customers’ instore movements by sequential pattern analysis using RFID data. These shopping paths are extracted using a pattern mining method. Finally, shopping paths are used in the decision tree analysis to generate the rules that expressed customers’ in-store movements and purchasing characteristics. As a result, in this study, the authors propose useful suggestions for more efficient in-store area management. movements and number of shelf visits. In recent years, with the decrease in the installation cost of RFID technology, it has been used not only in traffic problems (Ustundag & Cevikcan, 2008), logistics, traceability studies and merchandising in the distribution industry but also in understanding customers’ in-store movements in the retail industry. There have also been attempts to attach RFID tags to shopping carts to determine customers’ in-store movement behavior by collecting customer tracking data called RFID data (Hui, 2009; Larson, 2005; Yada, 2009). In this study, we identify customers’ shopping behavior by considering both in-store movements and purchasing behaviors using DOI: 10.4018/joci.2011100101 2 International Journal of Organizational and Collective Intelligence, 2(4), 1-14, October-December 2011 Copyright © 2011, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited. both RFID data and POS data with IDs obtained from a certain supermarket in Japan. At first, we classify customers into prime customers and general customers and extract the shopping path that expresses a sequence pattern obtained from a visiting order for the identification of each customer group. Analyzing RFID data by the sequences enables the evaluation of shopping behavior that was previously difficult: in what order customers buy things, under what circumstances they exhibit indecisiveness, etc. Finally, we use the extracted patterns to build a decision tree model and identify the characteristic shopping behavior of prime customers. The remainder of this paper is organized as follows. In the second section, we describe the related work about string analysis and shopping path analysis; then we show that the important factors are both the visiting order and classification problem using sequential patterns. In the third section, we introduce the purpose of this paper, definition of the object variable and detailed methods of this analysis. In the fourth section, we describe calculation experiments, which are extracted sequence patterns, and construct decision tree model. In the fifth section, we summarize the main results and introduce the future tasks.