A Comparative Analysis of Market Segmentation Using Machine Learning

Aug 3, 2026

Uploaded by Ophelia

Objective: This study compares behavioral customer segmentations for private-label products using machine-learning clustering algorithms. Retailers increasingly use private labels to balance competitive pricing with quality comparable to traditional brands. Two methods are applied: Simple K-Means (SKMA) and Expectation–Maximization Clustering (EMC). The research analyzes 1,073 customer loyalty surveys, builds an experimental framework from 23 selected questions, and implements both clustering approaches. Results reveal clear differences across four predefined clusters and indicate distinct potential marketing strategies.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages12

Reading Time30 min

AccessFree to download

Abstract

Objective

Theoretical framework

Method

Results and discussion

Research implications

Originality and value

A COMPARATIVE ANALYSIS OF MARKET SEGMENTATION USING MACHINE LEARNING a Carlos Hernandez, b Galo Paiva, c Magaly Sandoval ABSTRACT Objective: The objective of this study is to compare behavioral customer segmentations for private-label products using machine-learning clustering algorithms. Theoretical Framework: Retailers increasingly develop private labels to attract customers with competitive prices while maintaining quality standards comparable to traditional brands. To explore customer behavior in this context, two algorithms are applied: Simple K-Means (SKMA) and Expectation–Maximization Clustering (EMC) . Method: The research was conducted in four phases: analysis, design, construction, and validation and discussion. In the analysis phase, 1.073 customer loyalty surveys were examined. During the design phase, 23 questions were selected to build the experimental framework. In the construction phase, the clustering algorithms were implemented. Finally, the resulting segmentations were compared and evaluated. Results and Discussion: The results show clear differences between the two algorithms. With four predefined clusters, SKMA assigns customers in proportions of 43%, 26%, 15%, and 15%, whereas EMC produces segments of 18%, 23%, 31%, and 29% . Research Implications: These differences highlight important practical implications: SKMA and EMC generate distinct segmentations that may lead to different marketing strategies. Originality/Value: This study contributes to the literature by showing how machine-learning techniques can address the complexities of behavioral segmentation. Its relevance lies in offering marketing departments an effective and rapid approach to segment customers intoday’s competitive markets. Keywords: private labels, behavioral market segmentation, clustering algorithms, machine learning. Received: 9/24/2025 Accepted: 11/28/2025 DOI: [https://doi.org/10.55908/sdgs.v13i12.4579](https://doi.org/10.55908/sdgs.v13i12.4579) a PhD in Engineering, Universidad Católica de Temuco, Temuco, [Chile. E-mail: Carlos.hernandez.zavala@uct.cl](Chile. E-mail: Carlos.hernandez.zavala@uct.cl)b PhD in Engineering, Universidad de La Frontera, Temuco, Chile. E-mail: [galo.paiva@ufrontera.cl](galo.paiva@ufrontera.cl) c Msc in Engineering, Universidad de La Frontera, Temuco, [Chile. E-mail: magaly.sandoval@ufrontera.cl](Chile. E-mail: magaly.sandoval@ufrontera.cl) ANÁLISE COMPARATIVA DE SEGMENTAÇÃO DE CLIENTESAPLICANDO APRENDIZADO DE MÁQUINA RESUMO Objetivo: O objetivo deste estudo é comparar a segmentação comportamental de clientes para produtos de marca própria utilizando algoritmos de agrupamento baseados em aprendizagem de máquina. Referencial Teórico: As cadeias de supermercados geralmente têm marcas próprias para atrair clientes com preços competitivos, mantendo uma qualidade comparáveis aos das marcastradicionais. Nesse contexto, para segmentar os clientes de acordo com seu comportamento, foram aplicados dois algoritmos: Simple K-Means (SKMA) e Expectation–Maximization Clustering (EMC) . Método: A pesquisa foi realizada em quatro fases: análise, design, construção e discussão. Na fase de análise, foram examinadas 1.073 pesquisas de fidelização. Durante a fase de design, selecionaram-se 23 perguntas para construir o quadro experimental. Na fase de construção, implementaram-se os algoritmos de agrupamento. Finalmente, as segmentações resultantes foram comparadas e avaliadas. Resultados e Discussão: Os resultados mostram diferenças entre as segmentações produzidaspor ambos os algoritmos. Com quatro clusters predefinidos, o SKMA atribui os clientes nas proporções de 43%, 26%, 15% e 15%, enquanto o EMC gera segmentos de 18%, 23%, 31% e 29% . Implicações da Pesquisa: Essas diferenças evidenciam implicações práticas relevantes: SKMA e EMC produzem segmentações distintas que podem levar a estratégias de marketing diferentes, direcionando o esforço de vendas para segmentos distintos.

FAQs

What is the main objective of the study?

To compare behavioral customer segmentations for private-label products using machine-learning clustering algorithms.

Which clustering algorithms are used and how do their results differ?

The study applies Simple K-Means (SKMA) and Expectation–Maximization Clustering (EMC). With four predefined clusters, SKMA and EMC produce noticeably different segment proportions, leading to distinct segmentations.

What data and steps are used to build and validate the segmentation models?

The research uses 1,073 customer loyalty survey responses, selects 23 questions to form the experimental framework, implements the two clustering algorithms, and then compares and evaluates the resulting segmentations.

Aug 3, 2026

Uploaded by Ophelia