Skip to content

Model Comparison: LightGBM vs. XGBoost (Archived)


Note: This document is archived and no longer relevant to the final project outcome.

This document was used for a preliminary comparison between LightGBM and XGBoost. However, the final model selection process, detailed in the Model Selection Report, revealed that the XGBoost model suffered from significant overfitting and was dropped from consideration.

The key takeaway is that while both models are powerful, LightGBM proved to be more stable and reliable for this specific dataset and feature set. The final analysis confirmed that LightGBM was the superior choice, not just in performance, but in its ability to generalize without overfitting.

Please refer to the main Model Selection Report for the complete and accurate story of how the champion model was chosen.