Table 1 Optimal hyper-parameters after exhaustive grid search

Decision tree

Class weights

auto, 0 to 0.20 and 1 to 0.80, 0 to 0.10 and 1 to 0.90, 0 to 0.05 and 1 to 0.95

Maximum depth

1 to 10 (8)

Minimum samples split

2 to nVars+1 (18)

Maximum features

auto, sqrt, log2

Random forest

Number of estimators

1000

Class weights

auto, 0 to 0.20 and 1 to 0.80, 0 to 0.10 and 1 to 0.90, 0 to 0.05 and 1 to 0.95

Maximum depth

1 to 10 (9)

Minimum samples split

2 to nVars+1 (24)

Maximum features

auto, sqrt, log2

Random forest (full)

Number of estimators

1000

Class weights

auto, 0 to 0.20 and 1 to 0.80, 0 to 0.10 and 1 to 0.90, 0 to 0.05 and 1 to 0.95

Maximum depth

1 to 10 (1)

Minimum samples split

2 to nVars+1 (63)

Maximum features

auto, sqrt, log2

Gradient boosting

Number of estimators

1000

Maximum depth

1 to 10 (1)

Minimum samples split

2 to nVars+1 (9)

Maximum features

auto, sqrt, log2

Learning rate

0.1, 0.05, 0.02, 0.01

LDA

Number of components

None or 1 to nVars +1

QDA

Regularizing parameter

0 to 1 (0.89)

Linear SVM

Class weights

auto, 0 to 0.20 and 1 to 0.80, 0 to 0.10 and 1 to 0.90, 0 to 0.05 and 1 to 0.95

C

0.001, 0.01, 0.1, 1, 10, 100, 1000

Radial SVM

Class weights

auto, 0 to 0.20 and 1 to 0.80, 0 to 0.10 and 1 to 0.90, 0 to 0.05 and 1 to 0.95

C

0.001, 0.01, 0.1,1, 10, 100, 1000

Gamma

0.1, 0.01, 0.001, 0.0001

Polynomial SVM

Class weights

auto, 0 to 0.20 and 1 to 0.80, 0 to 0.10 and 1 to 0.90, 0 to 0.05 and 1 to 0.95

C

0.001, 0.01, 0.1,1, 10, 100, 1000

Gamma

0.1, 0.01, 0.001, 0.0001

Logistic regression

Class weights

auto, 0 to 0.20 and 1 to 0.80, 0 to 0.10 and 1 to 0.90, 0 to 0.05 and 1 to 0.95

C

0.001, 0.01, 0.1,1, 10, 100, 1000