Comparing Sentiment Analysis of Indonesian Presidential Election 2019 with Support Vector Machine and K-Nearest Neighbor Algorithm

Abstract

The purpose of this research is to compare the effectiveness of the Support Vector Machine (SVM) algorithm with the K-Nearest Neighbor (KNN) algorithm in predicting the results of the Indonesian presidential election 2019 based on sentiment analysis on social media (Twitter). The research methodology used includes several stages: preprocessing and weighting using TF-IDF. The SVM algorithm has the highest accuracy compared to the KNN algorithm. The average accuracy of SVM algorithm is 69.27, with the highest accuracy is 76.5%, while the average value of the KNN algorithm is 61.3% with the highest accuracy of 68.3%. The fastest training time is obtained by the KNN algorithm, while the SVM algorithm obtains the fastest testing time. The results of presidential predictions based on positive sentiment, namely candidate number 01 obtained a percentage of 67.98% while the number of positive sentiment predictions from candidate number 02 was 67.79%.