Strategic Customer Segmentation Using Unsupervised Learning and PCA: A Data-Driven Approach to Personalized Marketing التقسيم الاستراتيجي للعملاء باستخدام التعلم غير الخاضع للإشراف وتحليل المكونات الرئيسية (PCA): نهج قائم على البيانات للتسويق المُخصص

كمال, رامي; محمد, نوال

doi:10.21608/mauta.2025.410326.1007

	Strategic Customer Segmentation Using Unsupervised Learning and PCA: A Data-Driven Approach to Personalized Marketing التقسيم الاستراتيجي للعملاء باستخدام التعلم غير الخاضع للإشراف وتحليل المكونات الرئيسية (PCA): نهج قائم على البيانات للتسويق المُخصص
منارة الاسكندرية للعلوم التجارية
Volume 1, Issue 2, July 2025, Page 211-182 PDF (1.24 MB)
Document Type: المقالة الأصلية
DOI: 10.21608/mauta.2025.410326.1007
View on SCiNiTO
Authors
رامي كمال ¹; نوال محمد²
¹اكاديمية اسكندرية للادارة والمحاسبة
²اكاديميه الاسكندرية للادارة والمحاسبة
Abstract
In the era of data-driven decision-making, businesses must leverage advanced analytical approaches to better understand customer behavior and optimize marketing strategies. This study integrates unsupervised learning with Principal Component Analysis (PCA) to enhance customer segmentation. Using a real-world customer dataset of 2,240 records, the analysis underwent rigorous preprocessing and feature engineering to ensure data quality and representativeness. PCA reduced the feature space from 23 variables to 3 principal components, preserving 82% of the variance while simplifying complexity. Agglomerative Hierarchical Clustering was applied, with the Elbow Method confirming four optimal clusters. The results revealed distinct groups: (i) high-income, high-spending customers (35%), (ii) middle-income families with moderate engagement (28%), (iii) low-income, price-sensitive customers (25%), and (iv) disengaged, low-spending customers (12%). Clusters were further validated using Silhouette Scores, which improved by 18% compared to demographic-only segmentation. Compared with traditional methods such as K-Means and DBSCAN, the PCA-enhanced approach demonstrated superior interpretability and stability. Key findings highlight that high-value customers are less responsive to promotions despite their spending power, while price-sensitive groups show higher responsiveness to targeted deals. These insights suggest that personalized marketing, tailored to the behavioral and demographic profiles uncovered, can significantly improve customer engagement and resource allocation.
Keywords
Customer Segmentation; Unsupervised Learning; Agglomerative Clustering; Principal Component Analysis (PCA); Data-Driven Marketing


Statistics Article View: 125 PDF Download: 9