Behavioral Segmentation of Ride-Hailing Users Using the K-Means Clustering Algorithm

Abstract

The rapid growth of the ride-hailing industry has generated massive volumes of transactional data (Big Data), containing valuable information regarding user behavior patterns. One of the major challenges faced by companies is the difficulty of mapping user characteristics at an individual level, resulting in marketing strategies that are often generalized and insufficiently targeted. This study aims to segment ride-hailing service users using the K-Means Clustering algorithm to identify groups of user behavior based on the Frequency, Monetary, and Rating attributes. The research methodology follows the stages of the Knowledge Discovery in Databases (KDD) process, including data identification, data preprocessing, and result interpretation using Orange Data Mining software. The results indicate that four optimal user clusters were formed. Profile analysis identified four user segments: Loyal Customers (frequent users with high satisfaction), Thrifty Customers (cost-conscious yet satisfied users), Risky Customers (frequent users who are dissatisfied with the service), and Passive Customers (users with relatively low transaction activity). This study recommends segment-specific Customer Relationship Management (CRM) strategies, with particular emphasis on service recovery for high-risk users and retention programs for loyal users to maximize the company's profitability.