Binary Classification on Fraud (holdout test)
0.003BCELEBM
Evaluation Results
| Method | Links | |
|---|---|---|
| EBMType=Tree-based GAM2026.04 | 0.003 | |
| NAMArchitecture=MLP (64, 64, 32), Optimizer=AdamW2026.04 | 0.003 | |
| APLRType=Piecewise-linear GAM2026.04 | 0.003 | |
| ParamBoostImplementation=JAX2026.04 | 0.003 | |
| MGCVImplementation=PyMGCV, Bases=CubicSpline2026.04 | 0.004 | |
| Linear/Logistic regressionImplementation=scikit-learn2026.04 | 0.009 | |
| NBMArchitecture=100 shared basis functions, Optimizer=AdamW2026.04 | 0.012 |