Skip to main content

Table 5 PCA dimensionality reduction results via 10-fold cross-validation for H. sapiens

From: Comparative analysis and prediction of nucleosome positioning using integrative feature representation and machine learning algorithms

Method

Feature

Parameters

PCA%

ACC

Sn

Sp

MCC

AUC

SVM

FCGR

K = 1 + 2 + 4

0.85

0.8758

0.8966

0.8552

0.7528

0.9288

 

FCGR + DAC

K = 4, lag = 2

0.9

0.8749

0.8856

0.8644

0.7507

0.9314

 

FCGR + TAC

K = 4, lag = 2

0.85

0.8752

0.8878

0.8626

0.7513

0.9306

 

FCGR + DACC

K = 4, lag = 2

0.95

0.8410

0.8236

0.8583

0.6825

0.9138

 

FCGR + TACC

K = 4, lag = 2

0.95

0.8369

0.8170

0.8565

0.6749

0.9099

 

FCGR + PCPseDNC

K = 4, λ = 8, w = 0.5

0.88

0.8727

0.8896

0.8561

0.7463

0.9284

 

FCGR + PCPseTNC

K = 4, λ = 8, w = 0.5

0.9

0.8719

0.8878

0.8561

0.7444

0.9281

 

All features

 

0.95

0.7746

0.6388

0.9087

0.5698

0.8906

ELM

FCGR

K = 1 + 2 + 4

0.88

0.8428

0.8636

0.8222

0.6866

0.9075

 

FCGR + DAC

K = 4, lag = 2

0.88

0.8461

0.8724

0.8200

0.6936

0.9128

 

FCGR + TAC

K = 4, lag = 2

0.85

0.8469

0.8698

0.8244

0.6952

0.9129

 

FCGR + DACC

K = 4, lag = 2

0.9

0.8458

0.8763

0.8157

0.6936

0.9095

 

FCGR + TACC

K = 4, lag = 2

0.88

0.8439

0.8808

0.8074

0.6902

0.9072

 

FCGR + PCPseDNC

K = 4, λ = 8, w = 0.5

0.85

0.8454

0.8645

0.8265

0.6918

0.9107

 

FCGR + PCPseTNC

K = 4, λ = 8, w = 0.5

0.88

0.8437

0.8627

0.8248

0.6882

0.9069

 

All features

 

0.88

0.8447

0.8843

0.8057

0.6923

0.9118

XGBoost

FCGR

K = 1 + 2 + 4

0.95

0.8513

0.8733

0.8296

0.7037

0.9175

 

FCGR + DAC

K = 4, lag = 2

0.85

0.8537

0.8667

0.8409

0.7080

0.9172

 

FCGR + TAC

K = 4, lag = 2

0.95

0.8859

0.9045

0.8674

0.7725

0.9491

 

FCGR + DACC

K = 4, lag = 2

0.93

0.8364

0.8601

0.8130

0.6741

0.9014

 

FCGR + TACC

K = 4, lag = 2

0.95

0.8395

0.8667

0.8126

0.6805

0.9050

 

FCGR + PCPseDNC

K = 4, λ = 8, w = 0.5

0.95

0.8463

0.8711

0.8217

0.6937

0.9147

 

FCGR + PCPseTNC

K = 4, λ = 8, w = 0.5

0.93

0.8498

0.8645

0.8352

0.7003

0.9155

 

All features

 

0.95

0.8423

0.8729

0.8122

0.6864

0.9051

  1. “PCA%” means contributing rate of principal component
  2. Best values are in bold
Back to article page

BMC Bioinformatics

ISSN: 1471-2105

Contact us