Light gbm has the metric with log_loss for binary or multi classification.
is Random Forest also has the loss function with log_loss?


Yes, there are decision tree algorithms using this criterion, e.g. see C4.5 algorithm, and it is also used in random forest classifiers. See, for example, the random forest classifier scikit learn documentation:

criterion: string, optional (default=”gini”) The function to measure the quality of a split. Supported criteria are “gini” for the Gini impurity and “entropy” for the information gain. Note: this parameter is tree-specific.

Please note that usually it is referred to as cross-entropy or information gain rather than log loss.

For some background information you can give this and the links provided in the answers a read (although they may refer rather to decision trees): When should I use Gini Impurity as opposed to Information Gain?

  • $\begingroup$ My final submission is calculating with log-loss. so should I use Gini or information gain? $\endgroup$ – slowmonk Feb 11 '20 at 7:51
  • $\begingroup$ @slowmonk What do you mean by "my final submission" and what exactly are you doing? $\endgroup$ – Sammy Feb 11 '20 at 8:39
  • $\begingroup$ im doing like kaggle like competion . $\endgroup$ – slowmonk Feb 11 '20 at 8:50
  • $\begingroup$ @slowmonk I am not sure whether it makes a big difference (see for example this question). Alternatively, you could check in the forums of the respective competition what others do. In the meantime, if your question here has been answered please accept the answer so the system can close it. $\endgroup$ – Sammy Feb 12 '20 at 16:02

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.