Sentiment Analysis in Determining Service Performance between In-Driver and Gojek based on Public Opinion using the Naive Bayes Method

Abstract

An application-based transportation system connected to an internet connection that is widely discussed by the public is an online motorcycle taxi. One of the leading online motorcycle taxi companies in Indonesia is Gojek and In-Driver. Each customer has a different level of satisfaction with the services provided by Gojek and In-Driver Indonesia, so there are always pros and cons in the form of suggestions and complaints. Judging from the existing problems, a solution is needed in the form of analyzing the suggestions and complaints received by the company. The problem of classifying a sentiment sentence into certain classes can be solved by the Multinomial Naive Bayes Classifier method. The data used amounted to 1000 data and the data used were the first 500 data and the second 500 tweet data as a comparison value. The results of the gojek data calculation resulted in an accuracy value of 73%, precision of 72%, recall of 100%, and f1-score of 84%. The results of the indriver data calculation obtained an accuracy value of 85.71% accuracy, precision of 85.185%, recall of 95.833%, and f1-score of 90.196%. This proves that the Naïve Bayes classification algorithm is more suitable for use on a smaller amount of data.