Apply TF-IDF and LDA to Explore Topics and Related Trends in Electric Vehicle Reviews
摘要
As the importance of Electric Vehicles (EVs) in environmental protection and energy sectors increases, and with various countries implementing supportive policies, major automakers are launching a range of EVs to keep up with this trend. This research aims to analyze the core issues consumers focus on regarding EVs. Previous related studies often conducted experiments using qualitative analyses such as surveys or market research. This research employs a topic model LDA (Latent Dirichlet Allocation) combining TF-IDF to explore trends related to EVs, using quantitative analysis to replace the traditional qualitative approach, and supports the research findings with relevant data or events. The data for this research were sourced from professional reviews on Greencarreports and Electrek, both major American news websites reviewing EVs. Reviews were collected from January 2020 to November 2023 using a web crawler. Data preprocessing steps included cleaning, word segmentation, and stemming. Subsequently, TF-IDF was used for keyword filtering, allowing weighted allocation of vocabulary before inputting into the topic model. The experimental results analyzed three topics: Topic 1 concerns EV policies, Topic 2 pertains to the construction of charging stations, and Topic 3 relates to factory construction, capacity demands, and performance indicators. These findings align with the content of the International Energy Agency’s 2024 white paper “Trends in Electric Cars,” and subsequent news reports also corroborate the research method effectiveness. The analysis reveals that consumer concerns about EVs differ from those about traditional cars, where attention typically focuses on mileage, fuel consumption, and interior design. In contrast, for EVs, the focus is on policies, charging station construction, and capacity, which are more forward-looking topics. This research aids in understanding EV market trends and provides reference points for relevant companies and policy-making.