GUFRONI, Acep Irham and Purwanto, Purwanto and Farikhin, Farikhin (2025) PREDIKSI KEMAMPUAN AKADEMIK MAHASISWA MENGGUNAKAN MODEL EDUCATIONAL DATA MINING BERBASIS SUPERVISED LEARNING ALGORITHM PADA SELEKSI NASIONAL BERDASARKAN PRESTASI (SNBP) DI PERGURUAN TINGGI. Doctoral thesis, UNIVERSITAS DIPONEGORO.

	Text cover aja.pdf Download (317kB)
	Text Disertasi Acep - COVER s.d. ABSTRACT.pdf Restricted to Repository staff only Download (3MB)
	Text Disertasi Acep - Bab. 1.pdf Download (298kB)
	Text Disertasi Acep - Bab. 2.pdf Download (2MB)
	Text Disertasi Acep - Bab. 3.pdf Restricted to Repository staff only Download (1MB)
	Text Disertasi Acep - Bab. 4.pdf Restricted to Repository staff only Download (4MB)
	Text Disertasi Acep - Bab. 5.pdf Restricted to Repository staff only Download (282kB)
	Text Disertasi Acep - Daftar Pustaka.pdf Download (309kB)
	Text Disertasi Acep - Lampiran.pdf Restricted to Repository staff only Download (16MB)

Abstract

Perguruan tinggi saat ini dihadapkan pada tantangan dalam proses seleksi penerimaan mahasiswa baru yang harus dilakukan secara cermat untuk memastikan calon mahasiswa yang diterima memiliki kemampuan akademik yang selaras dengan bidang keilmuannya. Jalur penerimaan berbasis prestasi, yang kini dikenal sebagai Seleksi Nasional Berdasarkan Prestasi (SNBP), dilaksanakan menggunakan kriteria yang telah ditetapkan, dengan pembobotan kriteria menyesuaikan kebijakan internal masing-masing perguruan tinggi pada tahun berjalan. Namun, tidak jarang ditemukan ketidaksesuaian antara kemampuan mahasiswa yang diterima dengan kompetensi yang dibutuhkan di program studi pilihan, sehingga berpotensi meningkatkan risiko putus studi (drop out). Upaya pencegahan dapat dilakukan dengan memprediksi kemampuan akademik mahasiswa sejak tahap awal seleksi. Perbandingan antara kemampuan mahasiswa saat proses admisi dengan capaian akademik pada tahun pertama dapat menjadi acuan dalam menetapkan pembobotan kriteria yang paling relevan dan berpengaruh untuk periode seleksi berikutnya. Prediksi ini dapat dilakukan menggunakan pendekatan Educational Data Mining (EDM). Penelitian ini mengembangkan model EDM berbasis Supervised Learning Algorithm, meliputi Decision Tree (DT), Random Forest (RF), Support Vector Machine (SVM), dan Extreme Gradient Boosting (XGB). Data penelitian terdiri dari data admisi SNMPTN tahun 2018– 2022, SNBP tahun 2023, serta data capaian akademik mahasiswa di tahun pertama. Hasil pengujian menunjukkan bahwa model dengan algoritma XGB memberikan kinerja terbaik, dengan tingkat keberhasilan sebesar 94%.
Kata kunci: Prediksi Kemampuan Akademik, Educational Data Mining, Supervised Learning Algorithm, Decision Tree, Random Forest, Support Vector Machine, Extreme Gradient Boosting.

Universities today face challenges in conducting new student admissions that must be carried out carefully to ensure that accepted candidates possess academic abilities aligned with their chosen field of study. The merit-based admission pathway, currently known as the National Selection Based on Achievement (Seleksi Nasional Berdasarkan Prestasi or SNBP), is implemented using predetermined criteria, with weighting adjusted according to the internal policies of each university in the given year. However, there are often mismatches between the abilities of admitted students and the competencies required in their chosen program, which can increase the risk of dropout. One preventive effort is to predict students’ academic abilities as early as the admission stage. Comparing students’ abilities during the selection process with their academic achievements in the first year can serve as a basis for determining the most relevant and influential weighting criteria for subsequent admission periods. This prediction can be conducted using an Educational Data Mining (EDM) approach. This study develops an EDM model based on Supervised Learning Algorithms, including Decision Tree (DT), Random Forest (RF), Support Vector Machine (SVM), and Extreme Gradient Boosting (XGB). The dataset comprises admission data from SNMPTN (2018–2022), SNBP (2023), and students’ first-year academic performance data. The study results show that the XGB algorithm achieved the best performance, with 94% success rate.
Keywords: Academic Performance Prediction, Educational Data Mining, Supervised Learning Algorithm, Decision Tree, Random Forest, Support Vector Machine, Extreme Gradient Boosting.

Actions (login required)

View Item