Skip to content

model_training_comparison_report.md

Latest commit

 

History

History
27 lines (23 loc) · 1.6 KB

model_training_comparison_report.md

File metadata and controls

27 lines (23 loc) · 1.6 KB

Model Training Comparison Report

Scenario: Swiss+Knockout vs Including Grand Finals

Accuracy

  • Baseline (Swiss + Knockout): 0.6196
  • With Grand Finals: 0.6132
  • Difference: -0.0064

Strategic Feature Importance Change

Values indicate how much the model relies on specific hero attributes.

Feature Importance (Baseline) Importance (With GF) Change
Late_Power 0.0644 0.0647 0.0004
Mid_Power 0.0446 0.0514 0.0068
Early_Power 0.0395 0.0418 0.0023
Difficulty 0.0208 0.0204 -0.0004
Flex_Pick_Score 0.0187 0.0203 0.0015
Economy_Dependency 0.0180 0.0181 0.0001
Primary_Lane 0.0152 0.0158 0.0005
Hard_CC_Count 0.0145 0.0136 -0.0008
Escape_Reliability 0.0125 0.0122 -0.0003
Damage_Type 0.0071 0.0072 0.0001

Interpretation

  • Accuracy: Higher is better. A significant drop might indicate the new data conflicts with old patterns (meta shift).
  • Feature Importance: Changes here show if the model is prioritizing different aspects (e.g., Early Power vs Late Power) due to the new Knockout games.