Comparison of Classification Algorithms toward Sentiment Analysis on Claim of Old Age Guaranteed

Abstract

The pros and cons of the Old Age Security (JHT) claim rules contained in the Minister of Manpower Regulation No. 2 of 2022 are unavoidable in Indonesia, especially on Twitter social media. Of course, this is something that deserves to be analyzed as government policy makers so as not to make a wrong move, one of which is using sentiment analysis techniques. The stages of the research carried out include data collection, pre-processing, the model used, experimentation and model assessment, as well as validation and evaluation of results. The researcher crawled 1000 Indonesian-language tweets on the pros and cons of claims related to Old Age Security using the R programming language and carried out the Text Cleaning Process. In the next stage, pre-processing is carried out using Case Folding, Stemming, Tokenizing and Stopword Removal. This research resulted in the sentiments of the Indonesian people being very happy with the president's statement asking for the regulation of the Minister of Manpower No. 2 of 2022 to be revised, then tested using 3 classification algorithms and the application of Particle Swarm Optimization feature selection. The results of the sentiment testing of the JHT benefit claim analysis obtained the highest value on the Support Vector Machine algorithm with the addition of Particle Swarm Optimization, namely the accuracy value of 82.92%, precision of 84.53%, recall of 80.90% and AUC value of 0.892..

Keywords: Support Vector Machine, Sentiment Analysis, Old Age Guaranteed

sar 0.892