I have been tasked to explain the principle of the XGBoost algorithm to non-technical people (think 1-2 slides in a powerpoint presentation to upper management).

I am currently working with the original papers : here for the paper specific to XGBoost and here for the orignal boosting idea by Friedmann.

I have come up with a short explanation for the boosting principle : the model is built iteratively, by concentrating on where the previous model made errors. XGBoost take advantage of the tree architecture and tools (splitting, pruning) to do so.

What would you add to this short description ?

  • 2
    $\begingroup$ I would make one on trees and one on boosting, as visual as possible, maybe anchored in a concrete example that is relevant $\endgroup$ – BookYourLuck Aug 22 at 20:38

The xgboost docs come with a nice introduction and examples. If it is really only about a few slides, this may be a good start: https://xgboost.readthedocs.io/en/latest/tutorials/model.html


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.