Table 5 Top 10 model performances ranked by robustness.

From: Multi-scalar risk drivers for a heat vulnerability assessment framework using machine learning algorithms

Feature Set

Model Algorithm

\(R^{2}\) Train

\(R^{2}\) Val

\(R^{2}\) Test

\(R^{2}\) Robustness Avg

D: HVI + L1 + L2

XGBoost

1.0000

0.8831

0.8375

0.8923

C: HVI + L2

RandomForest

0.9840

0.8927

0.7932

0.8563

D: HVI + L1 + L2

RandomForest

0.9794

0.6723

0.7034

0.8439

D: HVI + L1 + L2

EBM-GAM

0.9924

0.7631

0.8406

0.7751

C: HVI + L2

LightGBM

0.9177

0.8413

0.6303

0.7683

B: HVI + L1

EBM-GAM

0.9812

0.8262

0.7864

0.7593

C: HVI + L2

EBM-GAM

0.9938

0.8443

0.6349

0.7427

D: HVI + L1 + L2

LightGBM

0.9594

0.7999

0.8613

0.7308

B: HVI + L1

RandomForest

0.9773

0.5605

0.5667

0.7152

B: HVI + L1

XGBoost

1.0000

0.5825

0.5590

0.7022

Back to article page