Skip to main content

Table 7 Runtimes (in seconds) of rule and tree learning methods on DOT and IDiscs datasets. The process of transforming original matrix onto ARFF file (build ARFF) and the process of building classification models were measured separately

From: Semantic biclustering for finding local, interpretable and predictive expression patterns

Split

DOT

IDiscs

 

Build ARFF

Build model

Test model

Build ARFF

Build model

Best model

  

J48

JRip

J48

JRip

 

J48

JRip

J48

JRip

1

1,033

1,237

26,810

17.00

23.44

274

59.59

510.84

3.08

3.11

2

1,091

1,503

21,384

19.45

18.67

272

38.03

557.92

2.93

3.19

3

1,042

1,076

19,519

19.09

18.19

287

71.62

363.00

3.16

3.16

4

1,096

1,300

20,054

17.59

19.07

270

64.65

438.87

3.16

3.25

5

1,127

2,010

20,605

18.61

21.22

278

39.47

941.30

3.20

3.64

6

1,121

1,999

24,568

19.38

18.69

260

39.77

550.50

3.11

3.05

7

1,097

1,656

25,279

18.90

18.60

281

47.61

288.14

2.98

3.00

8

1,058

1,087

22,459

26.47

18.48

269

44.00

641.16

3.14

3.26

9

1,023

1,236

14,062

17.81

18.24

288

54.83

201.10

3.25

2.91

10

1,268

1,583

27,299

18.81

21.07

276

42.83

506.14

2.96

3.06

\(\bar {x}\)

1,096

1,469

22,204

19.31

19.57

629.4

50.24

499.9

3.10

3.16

s d(x)

±70.6

±343

±3,995

±2.64

±1.75

±32.3

±11.78

±204.8

±0.11

±0.2