A Predictive YouTube Data Miner

This project aims to develop a YouTube data miner that will draw conclusions based on the openly available data available about a YouTube personality. Specifically it will analyze comments, upload history and subscriptions. It will use data about the upload history and try to determine the next upload date and time (as some YouTube personalities are sporadic). The agent will gather comments on the personality’s top videos and , from this, try to predict the like / dislike ratio for a new video based on the keywords associated with it. It will also try to predict the gain / loss ratio of subscribers created by new videos. Applications of this agent could be to maximize the amount of subscribers and minimize the ones who leave by formulating the gain/loss ratio. The application of the like / dislike prediction could help YouTubers to make the content they create better suited to the people who watch their videos. The application of the upload date determination will help those who are subscribed to the channel know when to expect a new video. With the application of these core features the agent will befit both those who upload videos and those who watch the videos that the content creators make.