Enhancing XGBoost Classification with SVM-SMOTE & EasyEnsemble for Imbalanced Telemedicine Sentiment Data

Abstract

Telemedicine is the practice of health through applications using audio, visual and data communication, including care, diagnosis, consultation and treatment as well as remote medical data exchange. Based on the results of sentiment analysis on telemedicine applications, imbalance data is often found. The purpose of this research is to identify the use of SVM-SMOTE and EasyEnsemble in improving the performance of XGBoost classification on sentiment data imbalance in Telemedicine. Identification is done by including SVM-SMOTE and EasyEnsemble methods in improving XGBoost Classification Performance using data obtained from the Halodoc application, then validation techniques will be carried out using AUC and GMeans. The results showed that the use of SVM SMOTE and EasyEnsamble for data imbalance in XGBoost obtained the best model that is feasible to use in improving the performance of imbalance classification of sentiment data in health applications.