Predicting Popularity of YouTube videos using Viewer Engagement Features

Online Video platforms like YouTube have become a huge part of people’s lives. Not just for entertainment, but as a medium of learning and expressing themselves. So, Popularity Prediction becomes important to support the development of such online video services to help content producers understand their viewers’ needs, likes, and dislikes. Here a model is presented that aims to predict and classify the popularity based on various viewer engagement features of a video like view count, like count and subscriber count. These features help us understand the viewer’s liking for certain content. So, a popularity measure has been introduced using these features and a classification model is used to classify the popularity as high, medium, or low. The highest results were obtained using the Random Forest classifier, which showed an accuracy of 91.12%.