IMPLEMENTASI SELF-ORGANIZING MAP (SOM) UNTUK KLASIFIKASI MINAT MAHASISWA TERHADAP PEMINATAN STUDI PRODI INFORMATIKA UNIVERSITAS XYZ

AYU PARNIDA SINAGA, . (2026) IMPLEMENTASI SELF-ORGANIZING MAP (SOM) UNTUK KLASIFIKASI MINAT MAHASISWA TERHADAP PEMINATAN STUDI PRODI INFORMATIKA UNIVERSITAS XYZ. Sarjana thesis, UNIVERSITAS NEGERI JAKARTA.

	Text COVER.pdf Download (591kB)
	Text BAB I.pdf Download (303kB)
	Text BAB II.pdf Restricted to Registered users only Download (683kB) | Request a copy
	Text BAB III.pdf Restricted to Registered users only Download (579kB) | Request a copy
	Text BAB IV.pdf Restricted to Registered users only Download (661kB) | Request a copy
	Text BAB V.pdf Restricted to Registered users only Download (186kB) | Request a copy
	Text DAFTAR PUSTAKA.pdf Download (220kB)
	Text LAMPIRAN.pdf Restricted to Registered users only Download (1MB) | Request a copy

Abstract

Pemilihan peminatan studi di Program Studi Informatika Universitas XYZ masih cenderung dipengaruhi oleh faktor subjektif tanpa mempertimbangkan rekam jejak akademik secara objektif, sehingga berpotensi menimbulkan ketidaksesuaian antara kemampuan akademik mahasiswa dan peminatan yang dipilih. Penelitian ini bertujuan menganalisis kinerja algoritma Self-Organizing Map (SOM) dalam mengklasifikasikan kecenderungan peminatan mahasiswa angkatan 2022 dan 2023 ke dalam tiga bidang peminatan, yaitu Rekayasa Perangkat Lunak (RPL), Teknik Komputer dan Jaringan (TKJ), dan Multimedia (MM), berdasarkan nilai mata kuliah dasar. Penelitian menggunakan metodologi Cross Industry Standard Process for Data Mining (CRISP-DM) dengan optimasi hyperparameter melalui grid search terhadap 960 kombinasi konfigurasi. Proses klasifikasi dilakukan melalui pelabelan neuron menggunakan metode majority voting. Kualitas peta SOM dievaluasi menggunakan Quantization Error (QE) dan Topographic Error (TE), sedangkan performa klasifikasi dievaluasi menggunakan confusion matrix dengan metrik accuracy, Precision, recall, dan F1-score. Analisis sensitivitas dilakukan dengan membandingkan model baseline dan model berbasis Synthetic Minority Oversampling Technique (SMOTE) untuk mengatasi ketidakseimbangan kelas. Hasil penelitian menunjukkan bahwa konfigurasi optimal model baseline dengan map size 20×20, sigma 3,0, learning rate 0,5, dan 5.000 iterasi menghasilkan akurasi sebesar 0,79, QE 0,0003, dan TE 0,0625. Penerapan SMOTE meningkatkan akurasi menjadi 0,83 serta meningkatkan recall peminatan MM dari 0,62 menjadi 1,00, meskipun recall TKJ menurun dari 0,67 menjadi 0,50 akibat heterogenitas karakteristik data pada peminatan tersebut. Hasil penelitian menunjukkan bahwa SOM efektif digunakan sebagai dasar sistem pendukung keputusan pemilihan peminatan studi. Selain itu, ditemukan diskrepansi sebesar 20,69% antara hasil klasifikasi dan peminatan yang dipilih mahasiswa, yang mengindikasikan adanya pengaruh faktor non-akademik dalam proses pengambilan keputusan peminatan. ***** The selection of study specializations in the Informatics Study Program at XYZ University is still largely influenced by subjective factors without adequately considering students’ academic records in an objective manner. This condition may lead to a mismatch between students’ academic competencies and their chosen specialization. This study aims to analyze the performance of the Self-Organizing Map (SOM) algorithm in classifying the specialization tendencies of students from the 2022 and 2023 cohorts into three specialization areas: Rekayasa Perangkat Lunak (RPL), Teknik Komputer dan Jaringan (TKJ), and Multimedia (MM), based on foundational course grades. The study employed the Cross Industry Standard Process for Data Mining (CRISP-DM) methodology, with hyperparameter optimization conducted through a grid search involving 960 configuration combinations. The classification process was carried out through neuron labeling using the majority voting method. SOM map quality was evaluated using Quantization Error (QE) and Topographic Error (TE), while classification performance was assessed using a confusion matrix with accuracy, Precision, recall, and F1-score metrics. Sensitivity analysis was performed by comparing the baseline model with a Synthetic Minority Oversampling Technique (SMOTE)-based model to address class imbalance issues. The results indicate that the optimal baseline SOM configuration, consisting of a 20×20 map size, sigma value of 3.0, learning rate of 0.5, and 5,000 iterations, achieved an accuracy of 0,79, with a QE of 0.0003 and a TE of 0.0625. The implementation of SMOTE improved the model accuracy to 0,83 and increased the recall for the Multimedia specialization from 0.62 to 1.00, although the recall for the TKJ specialization decreased from 0.67 to 0.50 due to the heterogeneous characteristics of the specialization data. The findings demonstrate that SOM is effective as a foundation for a decision support system in study specialization selection. Furthermore, a discrepancy of 20.69% was identified between the model classification results and the specializations chosen by students, indicating the influence of non-academic factors in the specialization decision-making process.

Item Type:	Thesis (Sarjana)
Additional Information:	1). Dr. Widodo, S.Kom., M.Kom. ; 2). Neng Ayu Herawati, S.Pd., M.T.
Subjects:	Sains > Matematika > Ilmu Komputer
Divisions:	FT > S1 Pendidikan Teknik Informatika Komputer
Depositing User:	Users 33922 not found.
Date Deposited:	23 Jul 2026 02:12
Last Modified:	23 Jul 2026 02:12
URI:	http://repository.unj.ac.id/id/eprint/66954

Actions (login required)

View Item