مقاله انگلیسی رایگان در مورد دسته بندی محصولات خرده فروشی بر اساس رفتار مشتری – الزویر 2017

Categorization is important in the retail business decision-making process. Product classification and customer segmentation belong to the most frequently used methods. The customer segmentation is focused on getting knowledge 5 about the structure of customers and is used for targeted marketing. For example [9] dealt with customer segmentation and its usability in marketing. Another approach to determining customer segmentation was used by [12]. Customer segmentation based on self-organizing maps with a priori prioritization in direct marketing was proposed in [19]. 10 The product categorization finds even more applications in marketing, e.g. new product development, optimizing placement of retail products on shelves, analysis of cannibalization between products and more general analysis of the affinity between products. A genetic algorithm to identify optimal new product position was proposed in [5]. A placement of retail products on shelves was 15 studied by [3]. Finding the right categories is also crucial for sales promotions planning. Cross-category sales promotion effect was studied in detail by [11] and [6]. Retail chains try to minimize costs everywhere. Among others, their aim is to minimize the costs of product storage in stores. The storage management 20 reaches the stage when stores often have no reserves in the drugstore storeroom because they are supplied dynamically two or more times per week. Therefore, it may happen that a store runs out of some products. The task is: 1. How to fill a free place on shelves until the storage is restored. 2. How to find a product that best substitutes for the original one. 25 Sold-out products are usually replaced by other ones from the same category, but it is not clear how to best define the categories from this viewpoint. This is the main business motivation behind this paper. Products are almost always categorized according to their purpose, package properties, e.g. package size, brand and price level. However, there are different 30 approaches to product categorization. For example [20] used hierarchical clustering while [24] promoted fuzzy clustering. Another interesting possibilistic approach to clustering both customers and products was published by [2]. Retail chains have available huge amount of market basket data, containing sets of items that a buyer acquires in one purchase, which can be used to ef35 ficiently model customer behaviour, e.g. [21]. However, these data are rarely taken into account in the product categorization. Data from market baskets are usually used for analysis of cross-category dependence for a priori given categories, e.g. [18], [15], [4] and [13]. This paper proposes a new method for choosing categories utilizing market 40 basket data. Our method classifies products into clusters according to their common occurrences in the shopping baskets. Sets of products in individual shopping baskets as they were registered by the receipts are the only data used by the method which assigns each product to just one category. The method determines product categories under given assumptions of product dependency 45 in the same category. It stems from the assumption that a customer buys only one product per category. Experience shows that customers who buy one product from a given category are generally less likely to buy also another product from the same category. The method applies a genetic algorithm to market basket data to find the best clusters of products based on their joint 50 occurrence in shopping baskets.