xgb.max_f1: xgboost evaluation metric for maximum F1 Score (Precision with Sensitivity harmonic mean)
Description
This function allows xgboost to use a custom thresholding method to maximize the F1 Score (Precision with Sensitivity harmonic mean). You can use this function via eval_metric. It leaks memory over time, but it can be reclaimed using gc().
Usage
xgb.max_f1(pred, dtrain)
Arguments
pred
Type: numeric. The predictions.
dtrain
Type: xgb.DMatrix. The training data.
Value
The maximum F1 Score (Precision with Sensitivity harmonic mean) for binary data.