We'll talk in more detail on Monday. A running tally of my thoughts on reading the methods/results.
I am confused why you are classifying user with the random forest algorithm - if the h-index already defines a user as expert or not, you don't need a machine learning model. You would only need the machine learning algorithm if you do not have an h-index score for a user. But isn't that information already defined in the dataset?
We'll talk in more detail on Monday. A running tally of my thoughts on reading the methods/results.