Skip to main content

Table 1 Regression benchmark performance

From: Synthetic learning machines

 

n

p

COBRA

RF

RFopt

SRF

Air

111

5

27.24

28.68

27.53

28.14

Air2

111

5

28.40

30.72

28.85

28.36

Automobile

193

29

9.83

8.94

6.79

7.52

Bodyfat

252

13

31.36

32.02

31.67

32.19

BostonHousing

506

13

18.88

14.64

12.39

12.80

BostonHousing2

506

16

17.44

13.57

11.32

11.61

CMB

899

4

96.33

100.90

90.32

89.86

Crime

47

15

61.74

59.99

59.51

59.03

Diabetes

442

10

57.58

53.22

53.14

55.20

DiabetesI

442

64

57.05

54.42

54.61

55.92

Fitness

31

6

83.34

66.48

59.61

57.76

Highway

39

11

38.84

43.67

33.95

32.18

Iowa

33

9

62.60

62.16

50.03

50.22

Ozone

203

12

26.90

26.19

26.20

26.42

OzoneI

203

134

27.42

26.14

26.32

26.08

Pollute

60

15

49.64

51.36

49.52

46.74

Prostate

97

8

87.32

46.02

46.95

50.12

Servo

167

19

15.22

21.47

11.27

11.99

ServoFactor

167

16

43.24

34.65

32.54

31.44

Tecator

215

22

13.84

16.11

13.48

6.19

Tecator2

215

100

31.24

34.21

30.64

27.94

Windmill

1114

12

31.64

31.39

31.31

32.15

expon

250

2

47.76

46.04

46.48

47.60

expon.noise

250

17

62.13

67.49

66.44

53.04

mlb.friedman1

250

10

21.46

26.11

24.15

19.04

mlb.friedman1.noise

250

10

30.91

34.77

33.13

30.48

mlb.friedman1.bigp

250

250

37.67

44.14

43.81

31.99

mlb.friedman2

250

4

13.94

14.75

14.24

14.04

mlb.friedman2.noise

250

4

37.19

36.77

36.80

38.58

mlb.friedman2.bigp

250

254

22.92

29.01

28.10

17.73

mlb.friedman3

250

4

19.21

22.01

19.87

15.59

mlb.friedman3.noise

250

4

37.47

39.38

38.53

36.97

mlb.friedman3.bigp

250

254

37.19

46.72

45.47

26.78

mlb.peak

250

20

14.75

17.24

16.28

6.21

mlb.peak.bigp

250

20

14.75

17.24

16.28

6.21

mlb.noise

250

500

101.69

100.75

100.47

100.29

sine

250

2

35.92

37.79

35.95

34.72

sine.noise

250

5

56.64

66.07

61.14

54.71

syn.ex1

250

50

20.69

30.87

28.57

8.54

syn.ex2

250

20

88.60

89.66

89.59

92.68

syn.ex3

250

50

43.88

47.88

47.50

43.04

syn.ex4

250

50

34.75

37.78

36.90

30.40

syn.ex5

250

20

62.50

65.07

64.82

62.80

syn.ex6

250

30

 

102.30

100.58

103.16

syn.ex7

250

300

55.13

61.68

61.38

52.41

syn.ex8

250

50

117.93

58.11

57.76

52.01

  1. Cross-validated and test-set standardized mean-squared error (MSE) performance over 100 independent replications. Standardized MSE obtained by dividing MSE by the variance of the Y-response and multiplying by 100.