Artificial Intelligence based Social Protest Effectiveness Analysis

Collective action has been employed across various historical contexts to influence societal change. Examples such as the suffragist and civil rights movements in the United States and recent farmers’ protests in Europe demonstrate its potential impact. However, predicting protest outcomes remains difficult due to the interaction of multiple factors. In this study, the factors associated with protest success are examined, and a machine learning approach is proposed to estimate their effectiveness. After data rebalancing, outlier removal, and hyperparameter tuning, the Random Forest model achieved 75% accuracy and a 59% F1 score on the Global Protest Tracker dataset. The proposed method is intended to support computational assessments of protest dynamics and to encourage collaboration between social and computational sciences.