Analisis Pemetaan Gaya Belajar Siswa Berdasarkan Model FSLSM Melalui Pendekatan Clustering Berbasis Densitas Dengan Reduksi Dimensi

Salsabilla, Rehana Putri (2026) Analisis Pemetaan Gaya Belajar Siswa Berdasarkan Model FSLSM Melalui Pendekatan Clustering Berbasis Densitas Dengan Reduksi Dimensi. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5027221015-Undergraduate_Thesis.pdf Restricted to Repository staff only Download (3MB) | Request a copy

Abstract

Perbedaan karakteristik gaya belajar siswa memengaruhi efektivitas proses pembelajaran, namun identifikasinya secara manual sulit dilakukan pada populasi yang besar dan beragam. Penelitian ini bertujuan mengembangkan sistem pendukung keputusan untuk pemetaan gaya belajar siswa berbasis Felder–Silverman Learning Style Model (FSLSM) menggunakan comparative clustering framework. Data diperoleh dari 245 siswa Bimbingan Belajar Genza Kauman melalui kuesioner Index of Learning Styles (ILS) yang diintegrasikan pada sistem berbasis web. Untuk mendukung evaluasi konsistensi model, dibentuk 490 data sintetis melalui augmentasi data sehingga diperoleh total 735 data sebagai pembanding. Penelitian membandingkan delapan model clustering, yaitu K-Means++, Fuzzy C-Means, DBSCAN, HDBSCAN, UMAP+K-Means++, UMAP+FCM, UMAP+DBSCAN, dan UMAP+HDBSCAN, menggunakan Silhouette Score (SI), Davies–Bouldin Index (DBI), Calinski–Harabasz Index (CHI), dan Composite Score sebagai metrik evaluasi. Hasil pengujian menunjukkan bahwa UMAP+HDBSCAN merupakan model terbaik dengan Composite Score 3,0000 pada kedua dataset. Pada dataset primer dihasilkan 22 cluster valid dengan SI=0,4945, DBI=0,6029, dan CHI=374,08, sedangkan pada dataset sintetis dihasilkan 62 cluster valid dengan SI=0,6284, DBI=0,4430, dan CHI=3259,42. Hasil clustering diimplementasikan ke dalam sistem pendukung keputusan berbasis web yang menyajikan profil gaya belajar siswa, karakteristik cluster, dan rekomendasi pembelajaran. Validasi pengguna yang melibatkan tiga tutor dan lima puluh siswa menunjukkan tingkat penerimaan sebesar 4,53 (90,7%) pada kelompok tutor dan 4,33 (86,6%) pada kelompok siswa. Hasil penelitian menunjukkan bahwa pendekatan comparative clustering framework berbasis UMAP dan HDBSCAN mampu memetakan karakteristik gaya belajar siswa secara efektif dan menghasilkan informasi yang relevan untuk mendukung pengambilan keputusan dalam proses pembelajaran. Sistem juga dilengkapi fitur pengelompokan dinamis berbasis pemotongan condensed tree HDBSCAN agar tutor dapat menyesuaikan jumlah kelompok belajar sesuai kapasitas operasional bimbingan belajar.
=================================================================================================================================
Differences in student learning style characteristics affect the effectiveness of the learning process, yet manual identification becomes challenging in large and diverse student populations. This study aims to develop a decision support system for mapping student learning styles based on the Felder–Silverman Learning Style Model (FSLSM) using a comparative clustering framework. Data were collected from 245 students at Genza Kauman Tutoring through an Index of Learning Styles (ILS) questionnaire integrated into a web-based system. To evaluate model consistency, 490 synthetic data instances were generated through data augmentation, resulting in a total dataset of 735 records for comparison. The study compared eight clustering models, namely K-Means++, Fuzzy C-Means (FCM), DBSCAN, HDBSCAN, UMAP+K-Means++, UMAP+FCM, UMAP+DBSCAN, and UMAP+HDBSCAN, using the Silhouette Score (SI), Davies–Bouldin Index (DBI), Calinski–Harabasz Index (CHI), and Composite Score as evaluation metrics. The experimental results showed that UMAP+HDBSCAN achieved the best performance, obtaining a Composite Score of 3.0000 on both datasets. In the primary dataset, the model produced 22 valid clusters with SI = 0.4945, DBI = 0.6029, and CHI = 374.08, while in the synthetic dataset it generated 62 valid clusters with SI = 0.6284, DBI = 0.4430, and CHI = 3259.42. The clustering results were implemented into a web-based decision support system that presents student learning style profiles, cluster characteristics, and personalized learning recommendations. User validation involving three tutors and fifty students resulted in an acceptance score of 4.53 (90.7%) among tutors and 4.33 (86.6%) among students. These findings demonstrate that the comparative clustering framework based on UMAP and HDBSCAN effectively maps student learning style characteristics and provides relevant information to support decision-making in the learning process. In addition, the system features dynamic grouping based on HDBSCAN condensed tree cutting, enabling tutors to adjust the number of study groups according to the operational capacity of the tutoring center.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Clustering, Learning Style, FSLSM, Conventional Clustering, UMAP, HDBSCAN, Decision Support System
Subjects:	T Technology > T Technology (General) > T57.5 Data Processing T Technology > T Technology (General) > T58.62 Decision support systems
Divisions:	Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Information Technology > 59201-(S1) Undergraduate Thesis
Depositing User:	Rehana Putri Salsabilla
Date Deposited:	31 Jul 2026 04:01
Last Modified:	31 Jul 2026 04:01
URI:	http://repository.its.ac.id/id/eprint/139540

Actions (login required)

View Item