This dataset has been created by scraping YouTube home page (so it contains suggested videos) where video thumbnails have been associated with number of views and title of the video. It could be a useful dataset for infering how much a good thumbnail and a good title drive the views of a video.