Segmenting the Retail Customers: A Multi-Model Approach of Clustering in Machine Learning

Segmenting the Retail Customers: A Multi-Model Approach of Clustering in Machine Learning

Mansurali Anifa, Mary Jeyanthi P., Dieu Hack-Polay, Ali B. Mahmoud, Nicholas Grigoriou
DOI: 10.4018/978-1-6684-4168-8.ch002
(Individual Chapters)
No Current Special Offers


The goal of “serving all” is similar to “serving none.” Marketers are constantly looking for ways to refine the way they segment markets. Segmentation involves diving markets into smaller portions (segments) of consumers with similar needs for a given good or service. This chapter explores the application of various algorithms and analytical techniques that are used to segment markets. These techniques include regression, cross-tabulation, hierarchical clustering, and k-means clustering performed through analytical tools such as R-Studio and MS Excel. The analyses drew upon the “customer data” dataset, which contained eight variables: age, income, marital status, ownership status, household size, family total sales, and family total visit. The findings demonstrate how such statistics could help the businesses understand the customers and target the specific customer with unique campaigns and offerings.
Chapter Preview


Segmentation is an integral part of marketing strategy. Accurate segmentation leads to better use of organisational resources and improves financial outcomes for a market-driven organisation. Kotler, Keller, Brady, Goodman and Hansen (2019) assert there is no single way to segment a market. A marketer must try different segmentation variables, in combination or alone. Catering to consumers’ different needs and preferences with a unique value proposition is mandatory for any market-driven organisation (Day, 2012).

Given the array of competing products found in most retail markets, segmentation helps marketers formulate and implement relevant strategies to promote, distribute, position, and price their goods and services. Organisations follow two major segmentation strategies: concentration and multi-segment strategies (Chinwendu, 2018). An organization will focus its marketing efforts on only one market segment using the concentration strategy. In the multi-segment strategy, a company focuses its marketing efforts on two or more market segments (Dibb & Simkin, 2016). Thus, companies are creating different marketing mixes for a different segment, and it is, therefore, necessary to define the segment and its characteristics. Market segmentation uses geographic, demographic, psychographic, and behavioural variables. The dataset considered in this study for the segmentation comprises marital status, age, income, marital status, owner status, family category, family value, household size. The data-driven approach using consumer data is explicitly critical for any business to succeed in its markets (Camilleri, 2020).

This tutorial study aims to demonstrate how retailers can improve the accuracy of market segmentation through the use of clustering techniques via machine learning. We use demographic variables acquired from Kaggle.

In machine learning, segmentation has been conducted using clustering techniques, an unsupervised learning method with known X, i.e. demographic variables, and an unknown Y— the segments to be concluded. Four segments are identified comprising consumer persona, which the company needs to know to devise the unique marketing strategies in terms of offers and offerings. Grouping customers into segments will help marketers understand their own customers' characteristics and competitors' customers. Further, segmentation addresses the homogeneity and heterogeneity of customers’ groups to help maximise the potential for a given good or service. Out of the various techniques available, our research has taken two a priori techniques, such as Pivoting and regression-based segmentation and two Post Hoc portioning based clustering methods, such as hierarchical and K –Means clustering.

Key Terms in this Chapter

Cluster Analysis: Is a statistical technique for data analysis. It operates by classifying objects into groups, or clusters, based on their degree of association. Clustering is an unsupervised learning approach, which means that prior to running the model, marketers have no idea how many clusters exist in the data.

Marketing Strategy: Is a plan of action in order to achieve a major or overall marketing aim ( Proctor, 2021 ).

Silhouette Algorithm: Offers a technique for interpreting and validating the consistency of data clusters—it generates a concise graphical depiction of the degree to which each object has been categorised properly ( Rousseeuw, 1987 ).

Marketing Analytics: Is a systematic process through which marketing business challenges are addressed via the use of data, statistics, mathematics, and technological innovation. Modelling and computer software is used to make marketing decisions.

Market Segmentation: The practice of dividing the market into distinct segments of customers with distinct requirements, desires, or characteristics—who may benefit from goods or services tailored specifically to them ( Grewal & Levy, 2020 ).

Complete Chapter List

Search this Book: