Klasifikasi Tingkat Depresi Menggunakan Algoritma Pohon Keputusan C4.5 Berbasis SMOTE-NC (Studi Kasus: Mahasiswa Fakultas Sains dan Analitika Data)

Inayatullah, Aisha Zahra (2026) Klasifikasi Tingkat Depresi Menggunakan Algoritma Pohon Keputusan C4.5 Berbasis SMOTE-NC (Studi Kasus: Mahasiswa Fakultas Sains dan Analitika Data). Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5003221184-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (9MB) | Request a copy

Abstract

Depresi merupakan salah satu masalah kesehatan mental yang banyak dialami mahasiswa dan dapat berdampak negatif terhadap prestasi akademik maupun kesejahteraan psikologis. Upaya deteksi dini diperlukan untuk mengidentifikasi mahasiswa yang berisiko mengalami depresi sehingga intervensi dapat dilakukan secara lebih tepat. Penelitian ini bertujuan mengklasifikasikan tingkat depresi mahasiswa Fakultas Sains dan Analitika Data (FSAD) Institut Teknologi Sepuluh Nopember (ITS) serta mengidentifikasi faktor-faktor yang paling berperan dalam klasifikasi tingkat depresi mahasiswa. Data penelitian diperoleh melalui survei terhadap 796 mahasiswa aktif FSAD ITS angkatan 2021–2024 dengan tingkat depresi yang diukur menggunakan instrumen Depression Anxiety Stress Scale-21 (DASS-21). Kategori depresi disederhanakan menjadi tiga kelas, yaitu Normal, Ringan, dan Berat. Model klasifikasi dibangun menggunakan algoritma Pohon Keputusan C4.5 berbasis Synthetic Minority Over-sampling Technique for Nominal and Continuous Data (SMOTE-NC). Untuk memperoleh konfigurasi model yang optimal, dilakukan evaluasi terhadap beberapa skenario pemodelan dengan kombinasi penggunaan variabel prediktor dan teknik penyeimbangan kelas. Optimasi hyperparameter max depth dilakukan menggunakan Grid Search dan Stratified 10-Fold Cross-Validation, kemudian model disederhanakan menggunakan Error-Based Pruning (EBP) dengan Confidence Factor sebesar 0,25. Hasil evaluasi menunjukkan bahwa model C4.5 berbasis SMOTE-NC menggunakan seluruh 13 variabel prediktor menghasilkan performa terbaik dengan nilai Accuracy sebesar 49,38% dan Macro F1-Score sebesar 0,4885. Model tersebut menghasilkan 41 aturan klasifikasi ‘If–Then’. Variabel Pemahaman Orang Tua terhadap Pola Pikir Anak merupakan faktor yang paling dominan dalam membedakan tingkat depresi mahasiswa dengan nilai Gain Ratio tertinggi sebesar 0,1451. Temuan ini menunjukkan bahwa faktor keluarga memiliki peran penting dalam klasifikasi tingkat depresi mahasiswa serta memberikan landasan bagi institusi pendidikan dalam merancang strategi deteksi dini dan intervensi kesehatan mental mahasiswa.
=======================================================================================================================================
Depression is one of the most prevalent mental health problems experienced by university students and may negatively affect both academic performance and psychological well-being. Early detection is therefore essential to identify students at risk of depression and facilitate timely intervention. This study aims to classify the depression levels of students at the Faculty of Science and Data Analytics (FSAD), Institut Teknologi Sepuluh Nopember (ITS), and to identify the factors that play the most significant role in the classification. The study used survey data collected from 796 active undergraduate students of the 2021–2024 cohorts, with depression levels measured using the Depression Anxiety Stress Scale-21 (DASS-21). The original depression categories were consolidated into three classes: Normal, Mild, and Severe. Classification models were developed using the C4.5 Decision Tree algorithm combined with the Synthetic Minority Over-sampling Technique for Nominal and Continuous Data (SMOTE-NC). To obtain the optimal model configuration, several modeling scenarios were evaluated based on combinations of predictor variables and class balancing techniques. Hyperparameter optimization of max_depth was performed using Grid Search and Stratified 10-Fold Cross-Validation, followed by model simplification through Error-Based Pruning (EBP) with a Confidence Factor of 0.25. The evaluation results showed that the C4.5 model with SMOTE-NC using all 13 predictor variables achieved the best performance, with an Accuracy of 49.38% and a Macro F1-Score of 0.4885. The model generated 41 ‘If–Then’ classification rules. Parental Understanding of Their Child's Mindset was identified as the most dominant factor in distinguishing students' depression levels, with the highest Gain Ratio value of 0.1451. These findings suggest that family-related factors play an important role in the classification of students' depression levels and provide a foundation for institutions to develop early detection strategies and mental health intervention programs in higher education settings.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	DASS-21, Depresi Mahasiswa, Ketidakseimbangan Kelas, Pohon Keputusan C4.5, SMOTE-NC. C4.5 Decision Tree, Class Imbalance, DASS-21, SMOTE-NC, Student Depression.
Subjects:	Q Science > Q Science (General) > Q325.5 Machine learning. Support vector machines. Q Science > QA Mathematics > QA76.9.D343 Data mining. Querying (Computer science) Q Science > QA Mathematics > QA278 Cluster Analysis. Multivariate analysis. Correspondence analysis (Statistics)
Divisions:	Faculty of Science and Data Analytics (SCIENTICS) > Statistics > 49201-(S1) Undergraduate Thesis
Depositing User:	Aisha Zahra Inayatullah
Date Deposited:	30 Jul 2026 03:47
Last Modified:	30 Jul 2026 03:47
URI:	http://repository.its.ac.id/id/eprint/139823

Actions (login required)

View Item