1

I was reading XGBoost's documentation here: https://xgboost.readthedocs.io/en/latest/tutorials/custom_metric_obj.html.

It's not immediately clear to me how the objective function and evaluation metric differs in their purpose. Which one is minimized in training? Are both used somehow?

desertnaut
  • 57,590
  • 26
  • 140
  • 166
rishai
  • 495
  • 3
  • 15
  • Despite the reference to Keras in the linked questions, the answers are in fact generally applicable, and clarify the differences between the objective function (loss) and the evaluation (or business) metrics, like the accuracy. – desertnaut Jul 27 '20 at 12:30
  • 1
    Yes, I believe the answers there explain the point well - thanks for pointing me to them. – rishai Jul 27 '20 at 12:33

0 Answers0