Analisis Sentimen Teehadap Mobil Listrik Menggunakan Indobert (Studi Kasus : Mobil Listrik Di Indonesia Pada Kanal Youtube)

Abstract

Electric vehicles (Electric Vehicles/EV) are currently becoming a favorite automotive product in Indonesia. Public interest in electric cars is quite high because they are considered more economical and environmentally friendly. However, the presence of this electric car has also caused controversy among the public. Based on dataindonesia.id, there are a number of obstacles for Indonesian people to adopt electric vehicles. One of the things that influences the level of adoption of electric vehicles in a country is public perception of electric cars. One of the media that is a forum for providing opinions on electric cars is youtube. This research analyzes electric car sentiment from various aspects obtained using K-means clustering. From the clustering results, two main clusters were obtained. The first cluster describes the battery aspect, while the second cluster highlights the price. Next, sentiment classification is carried out for each aspect using the BERT Indobenchmark/IndoBERT-base-pl model. In previous research, the Indobenchmark/IndoBERT-base-pl BERT model was concluded to have good performance in using Indonesian language text classification. The results of the multi-aspect sentiment classification analysis show that the sentiment classification accuracy for the battery aspect is 79% and the sentiment classification accuracy for the price aspect is 85%. Information obtained from sentiment analysis can be used to support strategic decision making related to product development, pricing and policy determination. Sentiment data can be a valuable input in the planning and decision-making process to increase electric car adoption.