Implementasi Metode Decission Tree Dalam Mengklasifikasi Depresi Menggunakan Rapidminer

Abstract

Depression has become a serious mental health problem with a significant impact on quality of life and work productivity. This study aims to develop a depression classification model using the Decision Tree method implemented through RapidMiner software. The dataset consists of 2054 data with 11 variables covering demographic aspects, working conditions, and mental health. Data preprocessing is carried out through several stages, including data format conversion, categorical variable transformation using Nominal to Binominal, and numeric data normalization with Z-transformation. Implementing the Decision Tree uses the gain ratio parameter as the criterion, maximum depth 10, and confidence 0.1, and activates the pruning and prepruning features for model optimization. The results showed excellent performance with an accuracy of 97.50%, a weighted mean recall of 92.29%, and a weighted mean precision of 93.49%. The confusion matrix shows that the model successfully identified 1463 non-depression cases and 139 depression cases correctly, with a low misclassification rate.