COMPARISON OF NAÏVE BAYES AND KNN ALGORITHMS IN THE ANALYSIS OF PUBLIC SENTIMENT REGARDING THE IMPLEMENTATION OF GOVERNMENT EMPLOYEES WITH TEACHER WORK AGREEMENTS

Abstract

Until now the implementation of education in Indonesia cannot be separated from the complexity of the management of educators, one of which is honorary teachers, and bureaucratic reform which has quite an impact on the quality of education and the work climate in it. As a way to increase public service satisfaction through the state civil apparatus (ASN), the Indonesian Ministry of Education and Culture agreed with the Ministry of Administrative Reform and Bureaucratic Reform and the Ministry of Finance to change the recruitment system or appointment of government employee teachers from the acceptance of prospective Civil Servants. Civil servants (CPNS) become government employees with a work agreement (PPPK) which in its implementation there are still obstacles and pros and cons, some agree and some disagree. therefore the researcher conducted research on sentiment analysis in data mining in the implementation of PPPK teachers on social media Twitter as many as 871 data which were then processed into 519 data. the author uses the Naïve Bayes and KNN techniques to determine the predictive effect of the Naïve Bayes and KNN algorithms on public opinion in the implementation of first aid instructors and compares the level of accuracy of the 2 methods. Researchers use the RapidMiner tool version nine.10.1. The prediction results of Naïve Bayes are 328 statistics with positive or agree sentiment and 191 information with negative sentiment, and the last is the prediction result from KNN, namely 315 information with positive sentiment and 204 facts with negative sentiment. analysis of public sentiment on the implementation of teacher first resource on social media Twitter with the Naïve Bayes algorithm achieves an accuracy of 75.53%. And the KNN achieves an accuracy of 73.41%. In this study, the Naïve Bayes method is a technique with a higher level of accuracy than the KNN with an accuracy rate of 75.53%.