Method for automatic classification of retail chain clients on the basis of clustering

Authors: Moskvichev N.V., Bekasov D.E.
Published in issue: #1(54)/2021
DOI: 10.18698/2541-8009-2021-1-665
Category: Informatics, Computer Engineering and Control | Chapter: System Analysis, Control, and Information Processing, Statistics
Keywords: retail, programming, neural network, Kohonen network, marketing, segmentation, classification, clustering
Published: 26.01.2021

The personalized approach to customers is necessary in modern retail and can be achieved by analyzing the customer base and providing each group of unique promotional offers. The article describes a method for automatic classification of clients based on clustering the client base using the Kohonen neural network. This method makes it possible to distinguish groups of customers by the similarity of their characteristics. The use of a neural network makes it possible to distribute customers into groups without a priori known exact number of groups, as well as to take into account the social-behavioral structure of customers for each store or network. The characteristics used in the RFM-analysis method, which are standard for marketing, are highlighted as the analyzed parameters. The allocated clusters are classified based on the average values of the characteristics of the clients of each cluster. Comparison with the RFM-analysis method is performed and the results of applying the developed method on real data are described. The presented method allows to reduce the number of considered groups of clients by more than 2 times. Its application can allow retail chains to adjust the supply more accurately in accordance with the nature of demand.

References

[1] Dolya oborota roznichnykh setey v obshchem ob’’eme oborota roznichnoy torgovli [Percentage of retail chains turnover in overall volume of retail turnover]. fedstat.ru: website (in Russ.). URL: https://www.fedstat.ru/indicator/40536 (accessed: 07.11.2020).

[2] Sysoeva S.V., Buzukova E.A. Kategoriynyy menedzhment. Kurs upravleniya assortimentom v roznitse [Category management. Course on assortment management in retail]. Sankt-Petersburg, Piter Publ., 2008 (in Russ.).

[3] Klassifikatsiya B2B i tipologiya klientov v B2C [B2B classification and B2C customer typology]. salesportal.ru: website (in Russ.). URL: https://salesportal.ru/statya-23-klassifikaciya-b2b-i-tipologiya-klientov-v-b2c (accessed: 07.11.2020).

[4] Loginova E.Yu. Marketing. Konspekt lektsiy [Marketing. Lecture notes]. marketing.wikireading.ru: website (in Russ.). URL: https://marketing.wikireading.ru/17167 (accessed: 07.11.2020).

[5] Segmentatsiya klientov po loyal’nosti ili RFM-analiz [Customer loyalty segmentation or RFM analysis]. loginom.ru: website (in Russ.). URL: https://loginom.ru/blog/rfm (accessed: 07.11.2020).

[6] Primenenie RFM-analiza v segmentirovanii klientskoy bazy [Using RFM analysis for segmentation of customer base]. habr.com: website (in Russ.). URL: https://habr.com/ru/company/unisender/blog/131225 (accessed: 07.07.2020).

[7] Tyurin, A.G., Zuev I.O. Cluster analysis, methods and algorithms of the clustering. Vestnik MGTU MIREA [Herald of MSTU MIREA], 2014, no. 2(3), pp. 86–97 (in Russ.).

[8] Gorbachenko V.I. Seti Kokhonena [Kohonen networks]. masters.donntu.org: website (in Russ.). URL: http://masters.donntu.org/2018/fknt/shumskyi/library/article8.pdf (accessed: 07.11.2020).

[9] Predobrabotka dannykh [Data pre-processing]. intuit.ru: website (in Russ.). URL: https://intuit.ru/studies/professional_skill_improvements/1609/courses/139/lecture/20569?page=3 (accessed: 07.11.2020).

[10] Aksenov, S.V., Novosel’tsev V.B. Organizatsiya i ispol’zovanie neyronnykh setey [Arrangement and application of neuron networks]. Tomsk, NTL Publ., 2006 (in Russ.).