I'm trying to come up with a good way to explain the 3 importance metrics (Gain, Cover, Frequency) to a layman with only a basic understanding of XGBoost and trees in general. How best would you frame this explanation?

  • $\begingroup$ Related: datascience.stackexchange.com/q/12318/29781 $\endgroup$
    – aivanov
    Commented Jan 8, 2018 at 15:35
  • $\begingroup$ Thanks aivanov, I saw that answer and I was wondeirn whether you can exapnd upon their definition of gain. "The Gain implies the relative contribution of the corresponding feature to the model calculated by taking each feature's contribution for each tree in the model. A higher value of this metric when compared to another feature implies it is more important for generating a prediction." How exactly is this calculated? $\endgroup$
    – ben890
    Commented Jan 9, 2018 at 1:44
  • $\begingroup$ While the suggested link is really detailed and nicely put, but here you are asking for layman's interpretation of feature importance, are not? Said that it might be hard to figure out the interpretation for each single importance metric! $\endgroup$ Commented Aug 26, 2018 at 19:40


Browse other questions tagged or ask your own question.