Table 3 Hyperparameter tuning of DES-MI(EIL) classifiers with bayesian optimization.

From: An interpretable dynamic ensemble selection multiclass imbalance approach with ensemble imbalance learning for predicting road crash injury severity

Approach

Hyperparameters ‘Optimal’ values

Region of Competence

k (2–20)

pct_accuracy

(0.6–0.9)

alpha

(0.5–1.0)

Instance Hardness

IH_Rate (0.1–0.5)

DES-MI (BRF)

7

0.82785

0.75752

0.40707

DES-MI (RBC)

10

0.83133

0.94815

0.22503

DES-MI (OBC)

6

0.68173

0.60519

0.17603

DES-MI (SPE)

7

0.70596

0.58351

0.32195

DES-MI

(BRF + RBC + OBC + SPE)

2

0.75556

0.90166

0.25776

Back to article page