Predicting Engagement of Brazilian Politicians on TikTok: A Machine Learning Approach

• Maria Santana UFRPE
• José Santana UFRPE
• Pablo Sampaio UFRPE
• Kellyton Brito UFRPE

Resumo

While established social media platforms like Facebook, Twitter, and Instagram have become staples in political campaigns, the 2022 Brazilian elections witnessed the rise of a new contender: TikTok. Despite its recent emergence in 2016, TikTok has already become the fourth most used social network in Brazil. This study investigates the potential of machine learning to predict engagement on the TikTok profiles of the two leading presidential candidates: Lula and Bolsonaro. Utilizing a dataset from previous studies, we implemented various machine learning models and found that the Support Vector Machine achieved the highest performance based on the F1-score metric for both candidates. Despite the results being better with Bolsonaro than with Lula, further analysis of metrics like recall and precision suggests valuable insights for social and political domains. These findings can aid both candidates and society in understanding what factors are most related to engagement on this emerging social media platform. Additionally, marketing and advertising teams can use this information to create content tailored to reach and engage with a politician’s target electorate.