Performance Improvement of Machine Learning Algorithm Using Ensemble Method on Text Mining

Ensembles are the concept of combining several algorithms into one. This concept is expected to overcome the weaknesses of each model and create more accurate and stable predictions in machine learning. The research conducted aims to compare the accuracy obtained from basic algorithms with ensemble machine learning. The ensemble learning used is stacking and voting. Both techniques are used because they have an influence on improving accuracy. In addition, the use of SMOTE is also to overcome the problem of data imbalance. The basic algorithms in this research use Random Forest, Support Vector Machine, Adaboost, and XGBoost. In general, this research was conducted to try to improve the accuracy of basic algorithms for sentiment analysis. The result of this research is the fact that the use of stacking and hard and soft voting can improve accuracy. The SMOTE method is also influential in improving accuracy.