Exploring Regression Models for Youtube Views Prediction with Interpretable Insights

This computer age has changed content consumption, every individual can access infinite collection of information at their fingertips. The change from conventional media to internet-based media has enabled consumers to select content, nurturing a varied and worldwide informational panorama and entertainment. The rise of trendsetters and community-driven content has more enhanced the effect of social media, generating new types of internet-based celebrity and modifying the way users and enterprises involve their viewers. YouTube has progressed into a massive video-sharing medium, presenting content covering from entertainment and tutorial to comprehensive informative materials. YouTube’s easy to use interface and computational suggestions produce customized viewing experience, contributing to platform’s accomplishment within its enormous user base. Revenue-generating initiatives on YouTube aid to a noteworthy financial source for content creators, cultivating the creation of varied and high-quality videos that captivate to both specific spectators and worldwide viewer base. The drive behind conducting view count prediction exists in its capability to enable content creators, marketers and platform planners. We use broad variety of attributes in this predictive analysis challenge to project the number of views on YouTube videos. The dataset is imported from kaggle. Features used are likes, view-velocity, comments, subscribers, total views, categories, tags.