Comparison of Decision Tree Algorithms and Support Vector Machine (SVM) In Depression Classification In Students

Abstract

Mental health in adolescents, especially students, is an important concern in the world of education. Early detection of symptoms of depression in students can help preventive efforts in handling them. This study aims to compare the performance of two classification algorithms, namely Decision Tree and Support Vector Machine (SVM) in detecting the level of depression in students based on data obtained from the Kaggle platform. The dataset used consisted of 502 student data with 10 features that caused depression and 1 target class. The research stage includes data preprocessing, which includes data cleaning, categorical value encoding, and normalization with the Min-Max Scaling method. The model was developed using the 5-Fold Cross Validation method to evaluate the classification performance of each algorithm. Model evaluation was carried out using precision, recall, and accuracy metrics. The test results showed that the SVM algorithm had better performance with a precision value of 93.63%, recall of 95.21%, accuracy of 94.22%, and F1-score of 94.68%. Meanwhile, Decision Tree obtained a precision of 81.77%, a recall of 84.90%, an accuracy of 82.86%, and an F1-score of 83.64%. Based on these results, it can be concluded that the Support Vector Machine is superior in classifying depression in students compared to Decision Tree