Analyzing the Online Reviews to Explore Recent Trends of the U.S. Automotive Industry by Latent Dirichlet Allocation Method

This study selects the U.S. automotive industry as the research subject to explore the recent trends in automotive development. Since the analysis was based on the content of reviews, a topic model for textual analysis is chosen as the methodology, with Latent Dirichlet Allocation (LDA) being the most representative topic model. A total of 7,492 online reviews of the U.S. automotive industry from the years 2020, 2021, and 2022 were collected. These data were then preprocessed to prepare them for input into the pre-set LDA model. The results of this study are as follows. In the market analysis, driving experience, comfort, interior, gas mileage, and other words for all years correlate with price/performance rankings in the U.S. automotive media. Based on the analysis of the 2020 Truck and SUV, the 2021 and 2022 SUV and Hybrid models, as well as the sales data of Tesla electric vehicles in 2022, a growing trend of Hybrid and pure electric vehicles could observe.