feat(phase2): LightGBM role-specific models - explored, v4 GBM retain… - #6
Open
Ajay03299 wants to merge 1 commit into
Open
feat(phase2): LightGBM role-specific models - explored, v4 GBM retain…#6Ajay03299 wants to merge 1 commit into
Ajay03299 wants to merge 1 commit into
Conversation
…ed as best predictor Trained separate LGBM (LightGBM) per role (BAT/BOWL/AR/WK) with Optuna 50-trial TPE tuning and 4-fold TimeSeriesSplit. Results vs v4 GBM baseline: MAE: 23.603 → 25.826 (+2.222 worse) Spearman/match: 0.1704 → 0.1142 (-0.056) Top-11 overlap: 52.89% → 51.64% (-1.25%) Pred std: 6.924 → 6.695 Root cause: splitting 20k rows into 4 role groups (~3-5k each) gives each model insufficient data. Single model benefits from cross-role pattern sharing. Cricket's irreducible variance (std=34) means R2~3-4% is the realistic ceiling for single-point regression regardless of architecture. Decision: retain v4 GBM predictions as backbone for Phase 3 ILP optimizer. The optimizer is the real differentiator - converting noisy predictions into constrained optimal team selection is the novel contribution.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
…ed as best predictor
Trained separate LGBM (LightGBM) per role (BAT/BOWL/AR/WK) with Optuna 50-trial TPE tuning and 4-fold TimeSeriesSplit. Results vs v4 GBM baseline:
MAE: 23.603 → 25.826 (+2.222 worse)
Spearman/match: 0.1704 → 0.1142 (-0.056)
Top-11 overlap: 52.89% → 51.64% (-1.25%)
Pred std: 6.924 → 6.695
Root cause: splitting 20k rows into 4 role groups (
3-5k each) gives each model insufficient data. Single model benefits from cross-role pattern sharing. Cricket's irreducible variance (std=34) means R23-4% is the realistic ceiling for single-point regression regardless of architecture.Decision: retain v4 GBM predictions as backbone for Phase 3 ILP optimizer. The optimizer is the real differentiator - converting noisy predictions into constrained optimal team selection is the novel contribution.