Model Klasifikasi Mental Siswa Menggunakan Algoritma Support Vector Machine

Abstract

Student mental health plays a vital role in academic performance and social well-being. This study aims to build a classification model using the Support Vector Machine (SVM) algorithm, based on 15 features covering demographic, academic, and behavioral aspects. The dataset, obtained from Kaggle, contains 426 records of junior and senior high school students. Key preprocessing steps include one-hot encoding, feature standardization, train-test splitting (80:20), and handling class imbalance with SMOTE. The model was trained using the Radial Basis Function (RBF) kernel and optimized using Grid Search CV to find the best parameters. Evaluation results show 65% accuracy, with better performance in predicting students without mental health issues (Absence). However, low recall for the Presence class indicates a need for improved strategies to handle data imbalance. This study highlights the potential of machine learning, particularly SVM, as a tool for early mental health detection in students, provided that effective data preprocessing is applied.