ANALISIS SENTIMEN ULASAN PLATFORM MERDEKA MENGAJAR (PMM) DENGAN ALGORITMA SUPPORT VECTOR MACHINE DAN NAÏVE BAYES

A breakthrough from government policy to overcome learning lag as a result of the Cov-19 pandemic with the Merdeka Teaching Platform (PMM) which can be downloaded on the Play Store. However, as technology develops, it will not run smoothly because of the many comments reviewed by users on the Play Store. So a sentiment analysis is needed to see the public's response to PMM. This algorithm can find out the community's response both positively and negatively using the SVM algorithm, Naïve Bayes, and the parameter tuning comparison algorithm. The data taken is only text data in the form of user comments on the PMM application Play Store with the range July 2022-May 2023. The testing process is carried out with 10% and 30% of the data set, and parameter tuning is carried out to get the best combination of parameters. As a result, PMM application users feel helped by this application because of its complete features, for example, there are student teaching materials and proof of work in line with the implementation of IKM, namely, teachers can easily collaborate through this application, while users who give negative reviews of this application say that the application PMM is not yet available in the form of an App Store or not yet available for IOS. From the two algorithms, it was found that both of them were able to provide good accuracy, namely above 80% with the SVM Algorithm having an accuracy of 92% and Naïve Bayes 90%, and the results of parameter tuning with the Best accuracy of 0.914 or 91.4%. Keyword: Analysis Sentiment, Naïve Bayes Classifier, Support Vector Machine, Tuning Parameter, Google Play Store