Table 1 Summary of the generalization performance of high-rank SNP features evaluated by lasso on the three datasets.

From: Elastic Correlation Adjusted Regression (ECAR) scores for high dimensional variable importance measuring

Data

Features’ number

ECAR

CAR

Lasso

Stability selection

SIS

Ridge

Base lasso

Spike

Length

(\({R}^{2}\hspace{0.17em}\)= 0.4,

\(\mathrm{\alpha }\hspace{0.17em}\hspace{0.17em}\)=  0.45)

5

342.9

357.3

361.1

331.4

331.2

366.1

272.8

10

320.4

330.1

339.0

317.1

325.0

332.2

272.8

20

303.6

308.7

314.9

301.1

317.8

317.0

272.8

30

301.9

306.2

310.2

296.3

316.1

307.2

272.8

Lodging

Degree

(\({R}^{2}\hspace{0.17em}\)= 0.32,

\(\mathrm{\alpha }\hspace{0.17em}\)= 0.4)

5

3.48

3.56

3.57

3.51

3.15

3.80

2.91

10

3.42

3.54

3.49

3.35

3.10

3.75

2.91

20

3.36

3.39

3.46

3.30

3.14

3.60

2.91

30

3.28

3.43

3.47

3.27

3.12

3.52

2.91

Leaf

Width

(\({R}^{2}\hspace{0.17em}\)= 0.45

\(\mathrm{\alpha }\hspace{0.17em}\)= 0.5)

5

0.067

0.067

0.070

0.062

0.060

0.069

0.050

10

0.061

0.061

0.063

0.059

0.056

0.066

0.050

20

0.058

0.058

0.059

0.056

0.055

0.062

0.050

30

0.056

0.056

0.059

0.055

0.059

0.060

0.050

  1. Base lasso is the prediction performance of lasso on the test sets using all features as input. See the “Methods” section for further details.