Closed hcho3 closed 7 months ago
Attention: 1 lines
in your changes are missing coverage. Please review.
Comparison is base (
5b1bbff
) 84.33% compared to head (c509bfa
) 84.29%.
Files | Patch % | Lines |
---|---|---|
src/gtil/predict.cc | 95.83% | 1 Missing :warning: |
:umbrella: View full report in Codecov by Sentry.
:loudspeaker: Have feedback on the report? Share it here.
When performing tree prediction in parallel, it is common to split the workload over the data row dimension. So it would be beneficial to put the row ID as the first (outer-most) dimension. In addition, both XGBoost and scikit-learn RandomForestRegressor puts the row ID in the first dimension when predicting with multi-target models.