---

name: cv-strategy description: Cross-validation configuration and fold management for this competition allowed-tools: Read, Grep, Glob

CV Strategy

Fold Configuration

CopyN_FOLDS = 5
SEED = 42

# Tabular
from sklearn.model_selection import StratifiedKFold
skf = StratifiedKFold(n_splits=N_FOLDS, shuffle=True, random_state=SEED)

# Image with groups
from sklearn.model_selection import StratifiedGroupKFold
sgkf = StratifiedGroupKFold(n_splits=N_FOLDS, shuffle=True, random_state=SEED)

Golden Rules

1. Same folds for ALL models - Required for proper stacking
2. No data leakage - Target encoding within fold only
3. Group awareness - Same source → same fold
4. Reproducibility - Always set random_state

Current Competition

• Competition: [Competition Name]
• Metric: [Evaluation Metric]
• Target: target column
• Groups: [Group column if applicable]

Fold Splits (Saved)

Copymodels/folds.csv
- fold_0: train=[...], val=[...]
- fold_1: train=[...], val=[...]
...

OOF Predictions

Copymodels/oof/
├── xgb_v1_oof.npy
├── lgb_v1_oof.npy
├── catboost_v1_oof.npy
└── efficientnet_b3_oof.npy

Best CV Scores

Leakage Checklist

• Target encoding uses train fold only
• Time-based features respect temporal order
• Group-based splits for related samples
• No test data in feature engineering