Clicks and Views - Analyzing YouTube Engagement Patterns Using Regression Modeling

This study investigates viewer engagement on YouTube using a comprehensive dataset containing over 575,000 video and channel entries. Key features such as likes/views, dislikes/views, views per subscriber, and channel activity are used to model and predict total view count and engagement behavior. The objective is to uncover patterns that contribute to content virality and user retention. Using linear regression via Scikit-learn, we identify significant predictors of channel view performance and evaluate model strength through R² and error metrics. Findings show that engagement ratios like dislikes/views and likes/views significantly impact visibility, suggesting that even controversial content can drive traffic. The research also employs visual analytics (scatter plots, heatmaps) to reinforce findings. Results suggest that digital visibility can be engineered through engagement optimization rather than content volume alone.