Optimization of Classification Models for Customer Sentiment on Train Suite Class Compartments Using SMOTE and Particle Swarm Optimization

This study uses three algorithms, namely Naive Bayes (NB), K-Nearest Neighbour (KNN), and Support Vector Machine (SVM). Then, the three methods are supplemented with the use of SMOTE (Synthetic Minority Oversampling Technique) and Particle Swarm Optimization (PSO), which will later be compared with the three methods to obtain good accuracy results. It is hoped that the use of SMOTE in this study can be a solution in handling imbalanced data, because the influence of imbalanced data is very large on the results of the model obtained, since algorithm processing that does not take into account data imbalance will tend to be dominated by the major class and ignore the minor class. Similarly, the use of Particle Swarm Optimization is expected to increase attribute weights and improve the accuracy of an algorithm and data classification. The model that obtained the best evaluation results was the Support Vector Model using SMOTE and Particle Swarm Optimization, with an accuracy value of 81.15%.