Table 5 Performance of models (G-mean)

original

0.692

0.800

0.771

0.458

0.586

0.543

0.528

0.532

0.565

0.854

MG

0.729

0.766

0.773

0.588

0.592

0.531

0.526

0.524

0.481

0.852

keywords selection (cut words: frequency < 2)

cubic interpolation

0.768

0.752

0.779

0.488

0.533

0.513

0.649

0.682

0.695

0.820

cut

0.758

0.798

0.808

0.402

0.594

0.601

0.530

0.532

0.565

0.849

mean

0.732

0.699

0.742

0.498

0.569

0.568

0.624

0.682

0.608

0.894

keywords selection (cut words: MI bottom 5%)

cubic interpolation

0.734

0.545

0.649

0.300

0.383

0.286

0.670

0.719

0.744

0.871

cut

0.690

0.569

0.697

0.522

0.546

0.460

0.317

0.397

0.574

0.860

mean

0.641

0.613

0.599

0.542

0.505

0.528

0.678

0.701

0.607

0.896

keywords selection (cut words: MI bottom 5% or frequency < 2)

cubic interpolation

0.715

0.628

0.599

0.372

0.392

0.439

0.667

0.687

0.757

0.851

cut

0.668

0.568

0.714

0.497

0.556

0.478

0.319

0.357

0.569

0.886

mean

0.714

0.599

0.605

0.488

0.555

0.560

0.676

0.675

0.609

0.900

keywords selection (cut words: frequency < 2) + MG

cubic interpolation

0.763

0.759

0.762

0.466

0.532

0.529

0.565

0.671

0.711

0.849

cut

0.725

0.723

0.818

0.482

0.576

0.528

0.526

0.526

0.483

0.845

mean

0.738

0.728

0.782

0.584

0.596

0.553

0.621

0.654

0.673

0.876

keywords selection (cut words: MI bottom 5%) + MG

cubic interpolation

0.698

0.606

0.620

0.222

0.000

0.282

0.661

0.728

0.754

0.864

cut

0.680

0.655

0.641

0.484

0.555

0.455

0.402

0.358

0.495

0.876

mean

0.697

0.621

0.590

0.490

0.501

0.477

0.656

0.642

0.766

0.894

keywords selection (cut words: MI bottom 5% or frequency < 2) + MG

cubic interpolation

0.703

0.604

0.581

0.293

0.000

0.405

0.677

0.722

0.716

0.845

cut

0.671

0.609

0.642

0.483

0.537

0.504

0.364

0.359

0.492

0.836

mean

0.697

0.621

0.590

0.478

0.563

0.529

0.678

0.647

0.666

0.884