Descriptive Analysis and Topic Modelling of X Posts to Detect Changes of Customer Trends in Pandemic Period: A Case Study of Jeans


Güneş Y., Arıkan M.

International Journal of Modern Research in Engineering and Technology, cilt.10, sa.8, ss.33-64, 2025 (Hakemli Dergi)

Özet

X (previously named Twitter), one of the social media platforms, is used for collecting data from customers and converting the data into valuable information for business. While X analyses are mostly conducted on current, political, cultural and daily issues that are of interest or rapidly spreading in society, sectoral analyses are conducted less frequently. The aim of the study is to reveal the effects of the pandemic on consumer trends and to extract the topics from consumers’ tweets. This paper deals with analyzing dataset of X messages related to jeans before pandemic and during the pandemic. The study focused on a niche topic like the clothing industry, through X by applying these three approaches: 1. A statistical and time analysis of descriptive features of tweets, 2. Topic modelling with words of context, 3. Revealing the change between pre-pandemic and pandemic period. Here, a total of 28 265 tweets posted between December 15, 2019 and December 31, 2020 was collected and processed for analysis. At the end of the study, it is determined that the most appropriate method for content detection of our dataset is unigram and LDA, and pre-pandemic agenda topics are categorized under 8 headings. Four new topics are added to these categories during the pandemic period.