Arysespajayadi Arysespajayadi, M. Djazman Addin Suryana, Andita Dani Achmad, Adlin Adlin, Ferawaty Syam

Strategi Promosi Penerimaan Mahasiswa Baru menggunakan metode DBSCAN

Introduction

Strategi promosi penerimaan mahasiswa baru menggunakan metode dbscan. Analisis strategi promosi penerimaan mahasiswa baru pakai DBSCAN. Identifikasi sumber informasi paling berpengaruh, seperti media sosial, untuk keputusan calon mahasiswa Politeknik Negeri.

Abstract

Penelitian ini mengimplementasikan algoritma DBSCAN (Density-Based Spatial Clustering of Applications with Noise) untuk menganalisis pola sumber informasi yang memengaruhi keputusan calon mahasiswa di Politeknik Negeri Media Kreatif. Data survei periode 2022–2024, yang terdiri dari 13 variabel, diproses menggunakan normalisasi min-max. Parameter DBSCAN ditentukan dengan nilai Eps 0,4 dan MinPts 3. Hasil analisis menghasilkan tiga klaster, dengan Klaster 1 (media sosial, internet, dan rekomendasi teman/sahabat) memiliki rata-rata normalisasi tertinggi (0,87), menunjukkan pengaruh signifikan sebagai sumber informasi promosi. Validasi menggunakan Skor Silhouette (0,31) dan Indeks Davies-Bouldin (0,66) menunjukkan kualitas klasterisasi yang moderat. Hasil ini menunjukkan bahwa algoritma DBSCAN mampu menangani data heterogen dari variabel sumber informasi promosi dengan cukup efektif. Namun, optimalisasi lebih lanjut dapat dilakukan untuk meningkatkan kohesi intra-klaster dan pemisahan antar-klaster

Review

This paper presents a timely and relevant application of the DBSCAN algorithm to understand the critical factors influencing new student admissions at Politeknik Negeri Media Kreatif. The objective to analyze patterns of information sources affecting prospective student decisions is highly pertinent for higher education institutions seeking to optimize their promotional strategies. By employing a data-driven approach using survey data from 2022–2024, the study demonstrates a commendable effort to move beyond anecdotal evidence and provide actionable insights into student recruitment dynamics. The methodology is clearly outlined, involving the processing of 13 variables using min-max normalization, followed by DBSCAN with specified parameters (Eps 0.4, MinPts 3). The core finding, the identification of three clusters, is particularly insightful. Cluster 1, comprising social media, internet, and friend/pal recommendations, is highlighted as having a significantly high normalized average (0.87), underscoring its pivotal role in influencing student choices. This result provides valuable empirical evidence that aligns with current trends in digital communication and peer influence, offering a solid foundation for targeted promotional campaigns. While the study successfully demonstrates DBSCAN's ability to handle heterogeneous data from promotional sources, the reported validation scores (Silhouette Score 0.31, Davies-Bouldin Index 0.66) suggest a moderate quality of clustering. As acknowledged by the authors, this indicates that there is significant scope for further optimization to enhance intra-cluster cohesion and inter-cluster separation. Future work could explore a more systematic approach to parameter tuning for DBSCAN, potentially through sensitivity analysis or expert-driven feature engineering, or consider comparative analysis with other clustering algorithms to identify the most robust segmentation. Despite this, the research makes a valuable contribution by applying advanced analytical techniques to a practical problem in higher education marketing.

Full Text

Comments

You need to be logged in to post a comment.