Hybrid approach for identifying strategic promotional locations using k-means clustering and support vector machine classification

Abstract

In the increasingly competitive landscape of higher education marketing, determining strategic promotional locations was essential to reaching prospective students effectively. This study proposed a hybrid machine learning framework combining K-Means clustering and Support Vector Machine (SVM) classification to identify high-potential areas for targeted promotional activities. The analysis used student enrolment data from 2021 to 2024, focusing on features such as city, province, and school origin. K-Means clustering was first applied to segment the data into three spatially and institutionally distinct clusters. These clusters were then used as pseudo-labels to train the SVM model, enabling the classification of new data points based on learned patterns. The model achieved a classification accuracy of 98%, with consistently high precision and recall across all clusters. Cluster interpretation revealed meaningful geographic and institutional differences that supported differentiated promotional strategies. Thematic map visualizations further enhanced the applicability of the model for geospatial decision-making. This study contributed to the development of data-driven, scalable, and interpretable solutions for location-based marketing. It also demonstrated the practical relevance of hybrid learning models in supporting strategic planning for educational institutions. Future work was suggested to incorporate additional socio-demographic variables and advanced ensemble methods to improve model robustness.