Analisis Sentimen Mengenai Subsidi Mobil Listrik Menggunakan Support Vector Machine berbasis Information Gain

Abstract

The use of electric cars in Indonesia continues to increase, with a percentage achievement of 7,679 units in 2022 from 229 units in 2020. The advantages of using electric vehicles include energy-saving, low noise levels, and low environmental pollution. In addition, the materials used for electric vehicles can come from various alternative energy sources, are easy to maintain, and can be recovered. recoverable. Because they use only batteries, electric cars save more money than regular fossil fuel vehicles. There are some people who argue that the use of electric cars can solve the gas oil shortage, but there are others who argue that the use of electric cars requires more careful planning, especially in terms of operational costs, infrastructure availability, and performance or speed. speed. This has prompted the government to create policies that regulate incentives for electric cars, as stated in Presidential Regulation No. 55 of 2019 on the Acceleration of Battery-Based Programs for Electric Cars. 2019 on the Acceleration of BatteryBased Programs for Road Transportation. The purpose of this research is to study public sentiment about the implementation of electric car subsidies taken from YouTube comment data. In this sentiment analysis, using the Information Gainbased SVM method was chosen because SVM can be used in handling complex data separations, especially when distinguishing between adjacent sentiment classes. SVM allows the formation of optimal decision boundaries for classifying sentiment while Information Gain is chosen because it selects the most informative features or those that provide the highest entropy reduction that can increase the performance of sentiment analysis. The results show that the use of Information Gain-based SVM method has high accuracy; SVM split data 20/80% obtained 0.78%, svm split data 10/90% obtained 0.79%, and Information Gain with split data 20/80% obtained 0.78%, and Information Gain with split data 10/90% obtained 0.80%.