Analisis Sentimen Gofood Berdasarkan Twitter Menggunakan Metode Naïve Bayes dan Support Vector Machine
Melati Indah Petiwi, Agung Triayudi, Ira Diana Sholihati
Abstract
The Covid-19 pandemic in Indonesia has an impact on every sector of life, including the economy. The government implements social activities that make people have to carry out activities at home. Because of this, humans choose to do everything digitally, including ordering food. With the application of public interest in ordering food online, the income of one of the food orders, namely Gojek (Gofood) has increased. However, Gofood has many pros and cons in the community. In this case, many people give their opinion about the use of social media, especially twitter. The purpose of this study was to analyze public opinion on the performance of Gojek (Gofood) in Indonesia. The grouping into three classes, namely positive, negative and neutral classes were tested using the Naïve Bayes and SVM methods and compared the two methods. The analysis of public sentiment regarding Gofood on Twitter resulted in 92.8% worthy neutral, 5.2% worthy positive and 2.0% worthy negative. Comparing the accuracy results, the Support Vector Machine method has greater accuracy than the Naïve Bayes method, with the Support Vector Machine accuracy values of 83% and 98.5%, while the Nave Bayes accuracy values are 74.6% and 91.5% respectively.