PERBANDINGAN METODE DECISION TREE DAN NAIVE BAYES CLASSIFIER PADA ANALISIS SENTIMEN PENGGUNA LAYANAN PT PERUSAHAAN LISTRIK NEGARA (PLN)

Abstract

Background : PLN is a state-owned company that is tasked with supplying electricity to all regions of Indonesia which certainly cannot be separated from the various obstacles experienced, to find out public sentiment on the services that have been provided, an analysis is carried out to determine public sentiment. The results of these sentiments are created in the dashboard using the Flask framework by comparing the Naive Bayes and Decision tree methods. To create a sentiment analysis dashboard for PT. PLN and make a research analysis model using a comparison of the Naive Bayes Classification and Decision tree methods. The method used in this research is Naive Bayes and Decision tree. The data obtained with a total of 40,745 Tweet data taken in the period 1 May 2022 - 4 June 2022 with the keyword "PLN". Making a dashboard that displays the results of the analysis where there is a menu to display the data and each analysis process. The use of 900 training data and 300 testing data resulted in the Naive Bayes method getting an accuracy of 83% on the training data and 80% for the Testing data, while the Decision tree method got an accuracy of 77% on the Training data and 56% on the Testing data. The analysis obtained for the method in this study also shows that the Naive Bayes method is better for classifying large amounts of data than the Decision tree. The sentiment generated by the highest number is negative, with most of the Tweets being complaints about the response to complaints and handling of damage reported by the public.