Using hybrid bayesian networks to detect audience behaviour changes in youtube
E.L. Castaño, Guillermo Leale · 2020
The nowadays volatile fame one could gain from the Internet led to a disruption in the media industry, with important repercussions in platforms such as the educational channels from YouTube. In this work, one of such channels, which was shut down after more than 2 years of activity, is studied through the use of a Hybrid Bayesian Network with Markov Chain Monte Carlo based sampling. With the application of our model, the behaviour of users can be inferred and thus find whether it changed at some point. As a result, it was indeed possible not only to identify two specific moments in time when that changed but also to provide a transition zone between the steady states before and after the change.