Comparative evaluation of hierarchical clustering and k-means for strategic customer segmentation in commercial companies

Abstract

In today's business environment, customer segmentation is a key resource for optimizing both decision-making and the design of business strategies based on data analysis. In this context, clustering techniques have gained relevance because they allow for the identification of behavioral patterns and the grouping of customers based on their consumption habits. This study aims to comparatively evaluate the performance of hierarchical clustering and K-means for the strategic segmentation of customers in companies in the retail sector, considering purchase frequency, transaction value, and customer tenure as variables. The research employed a quantitative approach based on data analytics, beginning with the preprocessing and normalization of the variables. Subsequently, the optimal number of clusters was determined using the elbow method. Hierarchical clustering was used for the exploratory diagnosis of the data structure, followed by the K-means algorithm for the final classification of homogeneous groups. Additionally, the Silhouette index was used as an internal validation metric to examine the quality of the resulting clusters. The results revealed the existence of customer segments with distinct behavioral patterns, distinguishing between frequent and occasional consumers. In conclusion, the combined implementation of these techniques enables the creation of reliable and consistent segmentations that support strategic decision-making and the development of personalized marketing tactics.