DETEKSI EMOSI PADA TEKS BERBAHASA INDONESIA MENGGUNAKAN PENDEKATAN ENSEMBLE

Emotion in written text is often difficult to recognize due to the absence of visual cues such as facial expressions or vocal intonation, which typically aid in understanding a person's feelings. This research aims to address this challenge by developing an emotion detection model for Indonesian text. The approach used is Ensemble Learning, combining three Machine Learning models: SVM, KNN, and XGBoost, to optimize emotion detection results. The main contribution of this research is the implementation of the Ensemble method for detecting emotions in Indonesian text, with performance evaluated using metrics such as accuracy, precision, recall, F1 score, and ROC AUC. The evaluation results show that the Ensemble model outperforms previous models, achieving an accuracy, precision, recall, and F1 score of 87.14%, and a ROC AUC score of 97.90%. To further enhance performance, this study utilizes GridSearchCV for hyperparameter tuning of the SVM and XGBoost models and employs the Automated Machine Learning (AutoML) tool TPOT to generate the KNN model.