Sahril Saputra, Kurniabudi, Jasmir

Analisis Perbandingan Algoritma K-Means Dan K-Medoids Dalam Mengukur Tingkat Kepuasan Mahasiswa Terhadap Pelayanan Akademik

Introduction

Analisis perbandingan algoritma k-means dan k-medoids dalam mengukur tingkat kepuasan mahasiswa terhadap pelayanan akademik . Analisis perbandingan algoritma K-Means & K-Medoids dalam mengukur kepuasan mahasiswa terhadap pelayanan akademik. K-Medoids (DBI 0.222) lebih optimal untuk clustering data.

Abstract

Penelitian ini bertujuan untuk menganalisis perbandingan algoritma K-Means dan K-Medoids dalam mengukur tingkat kepuasan mahasiswa terhadap pelayanan akademik di Institut Islam Mamba'ul Ulum Jambi. Data kepuasan mahasiswa dikumpulkan melalui kuesioner dan dianalisis menggunakan kedua algoritma tersebut dengan bantuan tools RapidMiner. Hasil clustering dievaluasi menggunakan Davies Bouldin Index (DBI) untuk menentukan algoritma yang paling optimal. Hasil penelitian menunjukkan bahwa mayoritas mahasiswa Institut Islam Mamba'ul Ulum Jambi merasa sangat puas dengan pelayanan akademik yang diberikan. Clustering dengan K-Means dan K-Medoids berhasil mengelompokkan mahasiswa ke dalam tiga cluster: "Sangat Puas", "Cukup Puas", dan "Tidak Puas". Algoritma K-Means menghasilkan cluster dengan jumlah anggota 450 ("Sangat Puas"), 351 ("Cukup Puas"), dan 218 ("Tidak Puas"). Sedangkan K-Medoids menghasilkan cluster dengan jumlah anggota 638 ("Sangat Puas"), 270 ("Cukup Puas"), dan 111 ("Tidak Puas"). Berdasarkan nilai DBI, algoritma K-Medoids (0.222) menunjukkan performa yang lebih baik dibandingkan K-Means (0.396) dalam mengelompokkan data kepuasan mahasiswa. Penelitian ini memberikan implikasi penting bagi Institut Islam Mamba'ul Ulum Jambi dalam mengevaluasi dan meningkatkan pelayanan akademik.

Review

This study, titled "Analisis Perbandingan Algoritma K-Means Dan K-Medoids Dalam Mengukur Tingkat Kepuasan Mahasiswa Terhadap Pelayanan Akademik," presents a focused investigation into applying data mining techniques to assess student satisfaction. The core objective is to compare the performance of K-Means and K-Medoids algorithms in clustering student feedback on academic services at Institut Islam Mamba'ul Ulum Jambi. By gathering data via questionnaires and utilizing RapidMiner for analysis, the research aims to identify the more effective algorithm for this specific application, with evaluation rigorously performed using the Davies Bouldin Index (DBI). The research successfully categorized student satisfaction into three distinct clusters: "Sangat Puas" (Very Satisfied), "Cukup Puas" (Moderately Satisfied), and "Tidak Puas" (Not Satisfied). A key finding indicates that a majority of students express high satisfaction with academic services. While both algorithms achieved this clustering, they produced different distributions within these groups. K-Means resulted in clusters of 450 ("Very Satisfied"), 351 ("Moderately Satisfied"), and 218 ("Not Satisfied") members. K-Medoids, on the other hand, yielded 638 ("Very Satisfied"), 270 ("Moderately Satisfied"), and 111 ("Not Satisfied") members. Critically, the DBI scores demonstrated K-Medoids (0.222) as superior to K-Means (0.396), signifying its better performance in creating more compact and well-separated clusters for this student satisfaction dataset. This research offers significant practical implications for Institut Islam Mamba'ul Ulum Jambi, providing a data-driven approach to evaluate and strategically enhance its academic services. The clear comparative analysis, supported by an objective evaluation metric like DBI, strengthens the validity of the findings regarding the optimal clustering algorithm for this context. While the abstract clearly outlines the methodology and key outcomes, further discussions in the full paper on the specific attributes of academic service measured, the rationale for selecting *k*=3, and how the characteristics of the "Not Satisfied" cluster specifically inform targeted improvement strategies would further enrich the contribution. Overall, the study delivers a valuable comparative analysis with clear, actionable relevance for institutional quality improvement.

Full Text

Comments

You need to be logged in to post a comment.