Cluster analysis of food retailers in Russia

Abstract

There is lack of studies on the theoretical and practical aspects of effective clustering of food retailers. The paper focuses on adapting the cluster analysis method to improve the financial performance of retail outlets by controlling relatively homogeneous objects, such as retail space, assortment depth, and average bill. Methodologically, the study relies on the theory of marketing. The research methodology rests on the adaptation of cluster analysis for food retailers. The information base includes retailers’ official websites, expert and analytical materials, as well as databases statista.com and 2gis.ru. The study presents the results of a competitive analysis of changes in the Russian retail market and identifies industry leaders and the most promising retail formats. We propose a mathematical model by using k-means clustering to calculate evaluation criteria and use them as the basis for building a ranking of a food retailer’s stores. The model was tested using the case study of a retail company in Perm (Perm krai, Russia). The identified evaluation criteria are sales volume, retail space, average bill, marginality, the number of SKUs, and service costs. The level of the dependence of retail development on these criteria is calculated. Based on the results of food stores clustering, we single out five clusters with similar approaches to the operational management of retail formats and determine the necessary inventory and logistics. The developed model of stores clustering contributes to the implementation of outlets provision standards and enhances retailers’ performance and the level of customer service.

Information technologies for solving problems of grouping objects based on cluster analysis methods

“…The difficulty of clustering processes lies in the fact that the entire set of objects can be divided into completely different clusters. This is due to the fact that the data describing the properties of objects can have a variety of types, usually represented as numeric and categorical [3,4] When performing cluster analysis, both hierarchical and non-hierarchical methods can be used. The main idea of hierarchical clustering is the sequential unification of smaller clusters into large (agglomerative methods) or the division of large clusters into smaller (divisive methods) [6].…”