The Sentiment Analysis of Online Customer Review on Food and Beverages Delivery Services in the GOJEK Application Using K-Nearest Neighbors

Abstract

The increasing of digitalization in developing an application of the public transportation service shows that the business is transforming rapidly and producing practicalities and innovation. For example, the use of the application has been advanced by adding delivery systems for the food and beverage businesses and many others. The massive number of users will give a massive amount of reviews about the applications. One of the companies that develop the application, such as Gojek, requires feedback from the user to improve its services for the future. The enormous reviews written by the users tend to be subjective, and difficult to interpret. By conducting analysis using the big data of reviews, Gojek can design the necessary strategies as a differentiator in attracting more customers to use its application. This study aims to further analyze the reviews based on a machine learning approach (K-Nearest Neighbors) and sentiment analysis to label each review as positive or negative. This study will use the best classification model determined by selecting the best accuracy and precision the model gives. The result gives that K-Nearest Neighbors get 83% accuracy on classified the reviews, and it tends to give a positive result with 35542 data. By analyzing the word frequency result, Gojek has succeeded in providing good application services and competent drivers.