Skip to main content

Table 5 F1-Score classification results of the recommended pipelines for the considered AeKNN architectures

From: Autoencoder-kNN meta-model based data characterization approach for an automated selection of AI algorithms

Dataset

AeKNN

(32)

(16)

(8)

(32,16,32)

(32,16,8,16,32)

APSFailure

0.9823

0.7553

0.9875

0.7573

0.9055

Higgs

0.8743

0.5451

0.5602

0.4938

0.5316

CustSat

0.9250

0.6366

0.4953

0.8194

0.5483

car

0.9635

0.9874

0.8144

0.7613

0.6817

kr-vs-kp

0.9246

0.7035

0.6532

0.5870

0.8751

airlines

0.5887

0.7928

0.5992

0.5707

0.3604

vehicle

0.8515

0.8204

0.2131

0.9099

0.3733

MiniBooNE

0.9715

0.9871

0.8873

0.7405

0.8531

jannis

0.7229

0.5748

0.8068

0.6911

0.6006

nomao

0.9343

0.9213

0.5395

0.8454

0.4294

Credi-g

0.9381

0.5772

0.5661

0.4141

0.5863

Kc1

0.9321

0.8389

0.9523

0.8583

0.4596

Cnae-9

0.8962

0.8741

0.6352

0.5938

0.7509

albert

0.8394

0.7036

0.6251

0.8074

0.9783

Numerai28.6

0.3747

0.5260

0.3029

0.4395

0.3540

segment

0.9130

0.8830

0.8837

0.7139

0.5426

Covertype

0.6886

0.6824

0.7249

0.4845

0.4620

KDDCup

0.9571

0.9974

0.7669

0.8386

0.7112

shuttle

0.9653

0.8537

0.4969

0.8306

0.7109

Gas_Sens-uci

0.6161

0.8660

0.9667

0.7667

0.8492

  1. The best ones are highlighted in bold