Predicting YouTube Video Popularity Using Machine Learning: A Comprehensive Analysis of Engagement Metrics

With the widespread adoption of social media platforms, user-generated content dominates online activities, and YouTube, as the most widely used video-sharing platform, continues to expand its influence. This study collected 52,867 videos published in 2024 from 370 YouTube channels, retaining 34,129 videos after data cleaning for analysis. The research employed various machine learning models, including Adaptive Boosting, Decision Tree, KNN, Multi-Layer Perceptron, Quadratic Discriminant Analysis, Random Forest, and Extreme Gradient Boosting, to investigate the predictive power of factors such as average view count, average comment count, average like count, video length, and publication time on video popularity. Results showed that the Extreme Gradient Boosting model performed best with 99.24% accuracy, while average daily views, comment count, and time since publication emerged as key predictors of video popularity. This study provides practical recommendations for content creators and marketers, emphasizing the importance of sustained audience engagement and interaction in enhancing video visibility and success rate.