PUBLIC SENTIMENT ANALYSIS OF 'DIRTY VOTE' DOCUMENTARY FILM ON TWITTER USING NAÏVE BAYES WITH GRID SEARCH OPTIMIZATION

The film "Dirty Vote" provides a realistic depiction of alleged fraud issues within Indonesia's democratic system, released ahead of the 2024 elections. This has sparked various public opinions, both in favor of and against the film, potentially affecting the stability of Indonesia’s democratic system. The aim of this research is to analyze the public's reaction to the "Dirty Vote" documentary, which could serve as a consideration for assessing public awareness in rationally responding to a film and improving the quality of democracy in Indonesia. This research will test the accuracy of data used in classification using the Naive Bayes Classifier based on collected Twitter data. The evaluation results of the Naive Bayes model for sentiment classification showed an accuracy of 86%, with a precision of 84% and a recall of 91%. When compared to the implementation of hyperparameter tuning using grid search with a stratified k-fold combination and parameter configurations for alpha: [0,1], binarize: [0.0], and fit prior: [true, false], better results were obtained with an accuracy of 90%, a precision of 87%, and a recall of 94%. This demonstrates that using parameter optimization methods from grid search can help improve the accuracy of a classification model. It is hoped that this research will contribute significantly to the development of Indonesia’s democratic system, particularly in raising public awareness to think more rationally and critically when evaluating and analyzing a film.