Prediction of Youth Political Involvement in Bangladesh through Social Media: A Machine Learning Perspective

Democratic growth nowadays depends on young political participation, especially in a technologically linked country like Bangladesh. Different social media platforms like Facebook, Instagram, TikTok, and X (Twitter) have developed into powerful tools for influencing political attitudes and motivating young people to participate civilly. This paper aims to forecast youth involvement in Bangladesh’s politics by means of social media interaction and machine learning techniques. A Google Form survey gathered a structured dataset including demographic, behavioral, and opinion-based aspects. Feature encoding and Synthetic Minority Oversampling Technique (SMOTE) handling of class imbalance were preprocessing tasks. Nine machine learning algorithms were evaluated, with CatBoost achieving the highest accuracy of 95.01%, followed closely by SVM and Random Forest. Explainable AI methods like LIME were employed to interpret the models’ predictions, highlighting key factors such as political campaign participation and social media influence. The results highlight the significant impact of social media on young political behavior and provide useful information for politicians and campaigners to use digital channels for youth participation. Deeper insights with larger datasets and advanced deep learning methods will be explored in future works.