Skip to main content
SpringerPlus

Table 6 Pairwise t tests on accuracy for different feature sets

From: Comparing writing style feature-based classification methods for estimating user reputations in social media

 

Base learners

C4.5

NN

SVM

NB

t

p

t

p

t

p

t

p

(a) Segmenting type s = like

 F1 < F1 + F2

3.1271

0.0038

3.9988

0.0004

8.3702

0.0000

3.6306

0.0010

 F1 + F2 < F1 + F2 + F3

0.0266

0.9790

−2.7879

0.0093

0.0410

0.9676

0.0000

1.0000

 F1 + F2 + F3 < F1 + F2 + F3 + F4

5.9653

0.0000

11.9349

0.0000

9.8836

0.0000

28.3710

0.0000

(b) Segmenting type s = dislike

 F1 < F1 + F2

1.1035

0.2744

0.7982

0.4280

2.3112

0.0245

−11.0785

0.0000

 F1 + F2 < F1 + F2 + F3

0.0000

1.0000

0.0369

0.9707

0.0000

1.0000

0.0000

1.0000

 F1 + F2 + F3 < F1 + F2 + F3 + F4

3.2182

0.0021

4.8950

0.0000

5.7765

0.0000

16.0591

0.0000

(c) Segmenting type s = sum

 F1 < F1 + F2

−0.5602

0.5775

1.7311

0.0888

5.8095

0.0000

−1.1181

0.2684

 F1 + F2 < F1 + F2 + F3

−0.0182

0.9855

0.0488

0.9612

−0.0824

0.9346

0.0000

1.0000

 F1 + F2 + F3 < F1 + F2 + F3 + F4

−9.4704

0.0000

−9.2471

0.0000

−0.5146

0.6088

−2.9386

0.0058

(d) Segmenting type s = portfolio

 F1 < F1 + F2

3.1271

0.0038

3.9988

0.0004

8.3702

0.0000

3.6306

0.0010

 F1 + F2 < F1 + F2 + F3

0.0266

0.9790

−2.7879

0.0093

0.0410

0.9676

0.0000

1.0000

 F1 + F2 + F3 < F1 + F2 + F3 + F4

5.9653

0.0000

11.9349

0.0000

9.8836

0.0000

28.3710

0.0000

 

Ensemble learning methods

RS-C4.5

RS-NN

RS-SVM

RS-NB

t

p

t

p

t

p

t

p

(a) Segmenting type s = like

 F1 < F1 + F2

11.8420

0.0000

11.7699

0.0000

18.7674

0.0000

17.8236

0.0000

 F1 + F2 < F1 + F2 + F3

0.9752

0.3370

0.4320

0.6687

−3.4808

0.0015

0.3547

0.7252

 F1 + F2 + F3 < F1 + F2 + F3 + F4

7.2221

0.0000

27.5745

0.0000

30.4662

0.0000

25.8677

0.0000

(b) Segmenting type s = dislike

 F1 < F1 + F2

2.3273

0.0235

3.6135

0.0006

3.8520

0.0003

7.6427

0.0000

 F1 + F2 < F1 + F2 + F3

−0.0326

0.9741

0.4753

0.6363

0.0316

0.9749

0.0000

1.0000

 F1 + F2 + F3 < F1 + F2 + F3 + F4

5.1154

0.0000

5.9364

0.0000

6.7400

0.0000

12.4236

0.0000

(c) Segmenting type s = sum

 F1 < F1 + F2

5.7001

0.0000

−4.7825

0.0000

9.0012

0.0000

27.3577

0.0000

 F1 + F2 < F1 + F2 + F3

−0.8536

0.3970

0.2574

0.7978

0.2793

0.7810

0.0000

1.0000

 F1 + F2 + F3 < F1 + F2 + F3 + F4

1.1808

0.2426

−15.1418

0.0000

9.2964

0.0000

20.1199

0.0000

(d) Segmenting type s = portfolio

 F1 < F1 + F2

11.8420

0.0000

11.7699

0.0000

18.7674

0.0000

17.8236

0.0000

 F1 + F2 < F1 + F2 + F3

0.9752

0.3370

0.4320

0.6687

−3.4808

0.0015

0.3547

0.7252

 F1 + F2 + F3 < F1 + F2 + F3 + F4

7.2221

0.0000

27.5745

0.0000

30.4662

0.0000

25.8677

0.0000

  1. The results are t and p values of the t tests for feature set comparisons, and the results more than 5 % of significance level are highlighted in italics
Back to article page