Evaluasi Performa Akademik Mahasiswa Kedokteran Menggunakan Metode K-Means Clustering

Andrian, Mochammad Dwiky (2023) Evaluasi Performa Akademik Mahasiswa Kedokteran Menggunakan Metode K-Means Clustering. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 05211940000012-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only until 1 September 2025. Download (4MB) | Request a copy

Abstract

Universitas dapat dikatakan berkualitas apabila memenuhi standar nasional perguruan tinggi yang dapat dilihat dari beberapa indikator, seperti nilai dan durasi studi dari mahasiswa. Pada program studi profesi dokter di Fakultas Kedokteran Universitas Muhammadiyah Surabaya (FKUM), ditemukan beberapa mahasiswa belum lulus ujian kompetensi pada percobaan pertama mereka, sehingga mereka harus mengulang kembali sampai lulus, dan hal tersebut dapat menambah durasi studi mahasiswa. Pada permasalahan tersebut, diperlukan evaluasi performa akademik mahasiswa program studi profesi dokter di FKUM. Analisis performa akademik mahasiswa dapat dilakukan melalui identifikasi cluster. Algoritma K-Means baik digunakan untuk melakukan clustering pada data akademik mahasiswa dan data cluster mahasiswa dapat membantu dosen wali dalam membimbing dan mengawasi proses belajar mahasiswa agar bisa lulus tepat waktu. Nilai k pada kumpulan data tertentu dapat ditentukan dengan pendekatan berdasarkan metode Elbow. Selanjutnya digunakan pendekatan algoritma K-Means++ untuk menentukan pusat cluster awal. Dengan metode ini diharapkan dapat melakukan identifikasi cluster dalam tujuan melakukan analisis dan evaluasi performa akademik mahasiswa. Dari K-Means Clustering, dihasilkan 8 cluster results dari 8 model yang berbeda. Pada evaluasi nilai SSE cenderung meningkat seiring bertambahnya fitur yang digunakan. Berdasarkan perhitungan Average, Standar Deviasi, ANOVA, dan Similarity Test, ditemukan bahwa Result 5 (Model 5) memiliki cluster result terbaik karena memiliki jumlah P-Value signifikan yang cukup banyak yaitu 11 (45.83%) dengan P-Value < 0.001 yang banyak yaitu 9 (37.50%), dan hanya dengan menggunakan 24 fitur, Result 5 dapat mewakili 114 fitur pada Result 8 dengan tingkat kemiripan sebesar 87%. Maka dari itu identifikasi karakteristik cluster berdasarkan Result 5. Identifikasi cluster menghasilkan tiga label cluster, yaitu mahasiswa dengan label Low Performance (16 mahasiswa) yang memiliki nilai rata-rata rendah, sedangkan mahasiswa dengan label Medium Performance (27 mahasiswa) memiliki nilai rata-rata sedang, dan mahasiswa dengan label High Performance (6 mahasiswa) memiliki nilai rata-rata tinggi. Karakteristik cluster menunjukkan bahwa mahasiswa dengan label Low Performance memiliki nilai rata-rata rendah pada Semester 1, Stase Ilmu Penyakit Dalam, Kulit, Pediatri, Radiologi, serta Ujian Kompetensi MCQ CBT – 1, mahasiswa dengan label Medium Performance memiliki nilai rata-rata sedang pada karakteristik yang sama, sedangkan mahasiswa dengan label High Performance memiliki nilai rata-rata tinggi. Hasil temuan tersebut memiliki implikasi penting dalam mengembangkan strategi pendekatan yang berbeda untuk setiap kelompok performa akademik mahasiswa
=====================================================================================================================================
A university can be considered of high quality if it meets the national standards for higher education, which can be assessed through various indicators such as students' grades and study duration. In the professional doctorate program at the Faculty of Medicine, Muhammadiyah University of Surabaya (FKUM), it was observed that some students did not pass the competency exam in their first attempt, leading them to repeat the exam until they pass, thus prolonging their study duration. This situation necessitates an evaluation of the academic performance of students in the professional doctorate program at FKUM. The analysis of students' academic performance can be accomplished through cluster identification. The K-Means algorithm proves to be effective in clustering academic data, and the resulting cluster data can assist academic advisors in guiding and monitoring students' learning processes to ensure timely graduation. The appropriate value of 'k' in a specific dataset can be determined using the Elbow method. Additionally, the K-Means++ algorithm is used to determine the initial cluster centers. These methods facilitate cluster identification for the purpose of analyzing and evaluating students' academic performance. Eight cluster results were generated from eight different models using K-Means Clustering. As the number of features used in the evaluation increased, the SSE value tended to rise. Based on the Average, Standard Deviation, ANOVA, and Similarity Test calculations, it was found that Result 5 (Model 5) exhibited the best cluster results, with a significant number of P-Values (45.83%) being less than 0.001, and a total of nine (37.50%) P-Values being less than 0.001. With only 24 features, Result 5 could represent 114 features in Result 8 with an 87% similarity rate. Consequently, cluster characteristics were identified based on Result 5. The cluster identification process yielded three cluster labels: Low Performance (16 students) with below-average grades, Medium Performance (27 students) with average grades, and High Performance (6 students) with above-average grades. Cluster characteristics indicated that students labeled as Low Performance scored low grades in Semester 1, the Department of Internal Medicine, Dermatology, Pediatrics, Radiology, and the MCQ CBT - 1 Competency Exam. In contrast, students labeled as Medium Performance had average scores in the same characteristics, while those labeled as High Performance achieved high scores. These findings have significant implications for developing different strategies to address the academic performance of each student group.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Academic Performance Evaluation, Data Mining, Clustering, K-Means.
Subjects:	H Social Sciences > HD Industries. Land use. Labor > HD38.7 Business intelligence. Trade secrets
Divisions:	Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Information System > 57201-(S1) Undergraduate Thesis
Depositing User:	Mochammad Dwiky Andrian
Date Deposited:	02 Aug 2023 02:25
Last Modified:	02 Aug 2023 02:25
URI:	http://repository.its.ac.id/id/eprint/101392

Actions (login required)

View Item