Pipeline mind map
Standard pipeline template for a other competition. Not yet specialised by an editor.
- 01Data processing
- 1.1Read the data page carefully
- 02Model design
- 2.1Simplest baseline first
- 03Training
- 3.1Local validation that mirrors the leaderboard
- 04Post-processing
- 4.1Match sample_submission
Metric
Roc Auc Score. ROC AUC: how well your scores rank positives above negatives (0.5 random, 1.0 perfect). Only order matters.
Public baselines
Most-voted public notebooks, refreshed daily from the Kaggle API (last 2026-11-16).
- S6E11 | Deep EDA & Artifacts | XGB OOF 0.950285 ▲73
- S6E11 | Two Families, One Blend | LB 0.95421 ▲62
- Credit Card Fraud | LB 0.9539 | Shallow-Tree Stack ▲62
- EDA | blend | LB-0.95239 | Credit Card Fraud ▲58
- Predicting Credit Card Fraud | Getting Started ▲56
- S6E11 Part 12 | Wide Stack + 2nd LM | LB 0.95298 ▲56
- S6E11 Fraud | GLM-Offset Trees & Review Budgets ▲50
- Seems Legit. Card Check: RF - AUC ▲39
- S6E11 LM Ladder | Public and CV Up | LB 0.95421 ▲38
- PS|S6|E11: RealMLP · PyTorch ▲36
- Fraud Signal Lab | GPT-2 and GLM Fusion ▲35
- Predicting Credit Card Fraud 101 ▲34