SmartYoutuber: A Data-driven Analytical Platform to Improve the Subscriber Growth and Sustainability using Artificial Intelligence and Big Data Analysis

Abstract

Youtuber is a new type of freelancer, whose economic profit and personal reputation are highly decided by their own popularity on the Internet, which can be reflected directly by the number of subscribers accumulated. In order to develop the management of YouTube channels and get more advertisement benefits, youtubers need to maintain the current subscriber group and appeal to more new followers by making attractive videos. But they lack efficient methods to analyze their video quality and their communication with subscribers so that they can predict their future development and adjust present strategies. In this paper, we applied several machine learning algorithm and models to study the prediction of short and long term future subscriber increase (we call them as growth and sustainability of youtubers) by analyzing youtuber-related information including video content(e.g. topic type, video tags, etc.) and subscriber interaction(e.g. views, likes, comments, etc.). One highest-scoring regression algorithm is proposed to make the out-performing prediction for certain youtubers, and we have proven its rationality and high accuracy in predicting the growth and sustainability of YouTube subscribers with suitable configuration. Apart from establishing algorithms, a relevant website, which offers services for future prediction and improvement suggestions, is created based on the established random forest regression algorithm. This application allows youtubers to completely analyze their current management situation and assists them to increase popularity for both social and economic benefits.