Descriptive data mining for multi-shelf product allocation in traditional retail

Abstract

The business expansion of minimarkets in small cities is one of the serious threats to the sustainability of traditional retail businesses or small independent retailers. Many traditional retailers eventually closed due to their inability to maintain competitiveness, as customers increasingly prefer shopping at modern retail outlets. A well-organized store layout can improve the shopping experience of customers, which has an impact on customer satisfaction and retail competitive advantage. Currently, shelf space allocation in traditional retail is still inattentive, making the placement of products on the shelf random and erratic. Based on these problems, this research aimed to design multi-shelf product allocation according to customer shopping patterns by combining clustering algorithms and market basket analysis (MBA). Clustering aims to divide data points into two different clusters, namely dominant product and less favored product, while MBA aims to identify the customer purchase pattern and preferences. The three MBA scenarios produced four, twelve, and forty rules. The research successfully designed two layouts by utilizing a combination of clustering and MBA algorithms. The utilization of data mining allows traditional retailers to extract information from the database to be arranged into a layout design that fits the shopping patterns and customer preferences.