CLASSIFICATION OF MYPERTAMINA APP REVIEWS USING SUPPORT VECTOR MACHINE

Abstract

Indonesia is rich in natural resources, including oil and gas, and it manages these strategic assets through state-owned enterprises, one of which is PT Pertamina. Pertamina is responsible for domestic fuel production, distribution, and price stabilization. To improve efficiency and transparency, Pertamina developed the MyPertamina application that enables cashless fuel purchases, stock monitoring, and up-to-date price information. The application aims to streamline distribution and control fuel prices, thus helping to stabilize the cost of goods and services. MyPertamina also ensures subsidized fuel distribution is more effective and targeted by identifying and verifying subsidy recipients, reducing the potential for abuse. A sentimental analysis of subsidized fuel user reviews using this application is needed to understand the public's views. This research uses the Support Vector Machine (SVM) method to analyze the sentiment of MyPertamina app reviews. This research produced a stable model. Out of 200 reviews, 190 were negative, and nine were positive, with an SVM model accuracy of 97%. Wordcloud visualization shows the words that appear frequently in each sentiment. Positive reviews appreciated the photo verification feature, easy payment, and good service. Negative reviews included verification difficulty, app error, and feature failure.