Sentiment and User Churn Analysis for Online Platform Business using Yelp Studies

In the digital era, online platforms are increasingly subjected to scrutiny with regard to user churn, which has a profound influence on their revenue, market share, and brand reputation. This paper examines restaurant user review data on Yelp and explores the relationship between users' review behaviour, different sentiment categories, and user churn. This paper presents a user stratification analysis and an engagement analysis to identify patterns among different user groups. The TextBlob tool was employed for sentiment analysis, with reviews categorised into five sentiment categories: highly positive, moderately positive, neutral, moderately negative, and highly negative. Subsequently, a method was established to calculate churn rates and a Chi-square test was conducted to explore the correlation between sentiment categories and user churn. Finally, a correlation analysis was conducted to examine the linear relationships between review characteristics and churn, thereby elucidating the determinants of user churn rate. The study reveals that elite users and the frequency of comments influence user engagement and churn. The Chi-square test results indicate a significant correlation between different user review sentiment categories and churn rates. Platforms can reduce user churn and retain elite users by organising contests, rewarding active users or introducing community-building features. These actions can lay a solid foundation for future online business success.