Prediksi Kelulusan Peserta Seleksi Nasional Berdasarkan Prestasi (SNBP) Menggunakan Pendekatan Deep Learning

Ramadhany, Tsabita Putri (2025) Prediksi Kelulusan Peserta Seleksi Nasional Berdasarkan Prestasi (SNBP) Menggunakan Pendekatan Deep Learning. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5025211130-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (21MB) | Request a copy

Abstract

Seleksi Nasional Berdasarkan Prestasi (SNBP) merupakan salah satu jalur penerimaan mahasiswa baru yang kompetitif di Indonesia dengan fokus utama pada nilai akademik dan prestasi siswa. Penelitian ini bertujuan mengembangkan sistem prediksi kelulusan peserta SNBP menggunakan pendekatan deep learning. Data yang digunakan merupakan data pendaftar SNBP di Institut Teknologi Sepuluh Nopember (ITS) pada tahun 2022 hingga 2024 yang mencakup nilai rapor, prestasi, dan pilihan program studi. Penelitian ini mengimplementasikan empat model deep learning, yaitu DNN, 1D-CNN, LSTM, dan GRU. Tiga skenario pengujian dilakukan secara berurutan, pertama membandingkan tiga variasi dataset untuk menemukan representasi data terbaik, kedua menguji pengaruh berbagai pendekatan penanganan data tidak seimbang melalui lima metode oversampling (SMOTE, Borderline-SMOTE, ADASYN, SMOTE-Tomek, SMOTE-ENN), sintesis data menggunakan CTAB-GAN, penerapan class weight, serta baseline tanpa penanganan apa pun, dan ketiga mengeksplorasi konfigurasi parameter model untuk meningkatkan performa prediksi. Hasil evaluasi menunjukkan bahwa skenario terbaik diperoleh dari penggunaan dataset dengan pembobotan tanpa mempertimbangkan akreditasi sekolah, tanpa penanganan data tidak seimbang, dan model DNN dengan konfigurasi optimal berupa satu dense layer berukuran 64, fungsi aktivasi ELU, optimizer RMSprop, dropout 0,6, tanpa batch normalization, serta batch size 16 yang menghasilkan F1-score sebesar 0,8554. Analisis menunjukkan bahwa fitur yang paling berpengaruh terhadap prediksi adalah pilihan program studi, nilai rapor pada mata pelajaran tertentu, prestasi yang diraih siswa, serta nilai mata pelajaran pendukung.
=================================================================================================================================================
University admissions in Indonesia are carried out through several pathways. One of the most competitive achievement-based pathways is the Seleksi Nasional Berdasarkan Prestasi (SNBP) which focuses on academic performance and student achievements. This study aims to develop a prediction system for SNBP admission outcomes using a deep learning approach. The dataset consists of SNBP applicants to Institut Teknologi Sepuluh Nopember (ITS) from 2022 to 2024 and includes report card grades, achievements, and program study preferences. Four deep learning models were implemented in this research: DNN, 1D-CNN, LSTM, and GRU. Three experimental scenarios were conducted sequentially. The first compares three dataset variations to find the best data representation, the second examines the impact of different imbalance handling approaches using five oversampling methods (SMOTE, Borderline-SMOTE, ADASYN, SMOTE-Tomek, SMOTE-ENN), data synthesis with CTAB-GAN, class weight application, and a baseline without any handling, and the third explores model parameter configurations to enhance predictive performance. The evaluation results show that the best scenario is obtained using a dataset with weighting that excludes school accreditation, without any imbalance handling, and a DNN model with an optimal configuration consisting of one dense layer with 64 units, ELU activation function, RMSprop optimizer, a dropout rate of 0.6, no batch normalization, and a batch size of 16, resulting in an F1-score of 0.8554. Further analysis indicates that the most influential features in the prediction are program study preferences, report card grades in specific subjects, student achievements, and supporting subject scores.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	SNBP, Prediksi Kelulusan, Deep Learning, Ketidakseimbangan Data, SNBP, Selection Outcome Prediction, Deep Learning, Data Imbalance.
Subjects:	Q Science > Q Science (General) > Q325.5 Machine learning. Support vector machines. Q Science > QA Mathematics > QA336 Artificial Intelligence Q Science > QA Mathematics > QA76 Computer software Q Science > QA Mathematics > QA76.6 Computer programming. Q Science > QA Mathematics > QA76.76.E95 Expert systems Q Science > QA Mathematics > QA76.87 Neural networks (Computer Science) Q Science > QA Mathematics > QA76.9.D343 Data mining. Querying (Computer science)
Divisions:	Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Informatics Engineering > 55201-(S1) Undergraduate Thesis
Depositing User:	Tsabita Putri Ramadhany
Date Deposited:	31 Jul 2025 02:52
Last Modified:	31 Jul 2025 02:52
URI:	http://repository.its.ac.id/id/eprint/124307

Actions (login required)

View Item