Abstract
This paper explores the application of supervised machine learning models to predict political behavior in social media contexts based on demographic characteristics and platform usage patterns. Using a structured survey dataset collected from Kazakhstani respondents prior to national elections, we define political behavior as a composite of actions such as voting intention, political content sharing, and trust in digital information. Features include region, gender, frequency of social media use, and preference for specific platforms (e.g., Telegram, Instagram, Facebook). Several models-including logistic regression, random forest, and gradient boosting-are trained to classify users by political engagement level. Evaluation metrics such as accuracy, F1-score, and ROC-AUC demonstrate the models’ ability to uncover latent correlations between user attributes and their digital political actions. Feature importance analysis further reveals which social and technological factors are most predictive of online political participation. The study contributes to the field of computational social science by illustrating how predictive modeling can supplement traditional survey-based insights, especially in contexts of low-resource languages and developing digital cultures.