Improving SVM Performance in Determining Cost and Kernel Parameter Values Through Grid Search

The purpose of this study is to improve the performance of Support Vector Machine (SVM) algorithm in sentiment analysis of trainee reviews through parameter optimization using Grid Search. Trainee reviews were taken from the Vocational and Productivity Training Center (BBPVP) Medan, which is the context of this study. The method used includes collection and preprocessing of review data, followed by the application of SVM for sentiment analysis. Grid Search is used to optimize the cost and kernel parameters of SVM. The results show that after optimization, the resulting SVM model shows significant performance improvement compared to the model before optimization. Accuracy increased from 68.75% to 71.96%, with improvements in other metrics such as precision, recall, and F1 score. These findings suggest that Grid Search is an effective method to optimize SVM parameters in sentiment analysis, providing a more accurate evaluation of the effectiveness of training programs at BBPVP Medan. This research makes an important contribution by applying hyperparameter optimization to SVM and suggests the application of this method for similar analysis in the future.