Skip to main content

Table 7 R 2 performance on test set 1

From: Multi-model inference using mixed effects from a linear regression based genetic algorithm

Variable selection

LASSO

GA-OLS

GA-MM

Variable estimation

Coef (shrinkage)

OLS

MM

OLS

MM

MMI1

MMI2

OLS

MM

MMI1

MMI2

      

OLS

MM

OLS

MM

  

OLS

MM

OLS

MM

TOP15 variables

0.815

0.816

0.827

0.830

0.833

0.832

0.838

0.827

0.831

0.834

0.835

0.834

0.839

0.829

0.832

TOP18 variables

0.816

0.818

0.827

0.830

0.831

0.833

0.838

0.825

0.829

0.832

0.835

0.835

0.839

0.829

0.832

TOP21 variables

0.819

0.825

0.835

0.821

0.825

0.836

0.839

0.819

0.824

0.824

0.826

0.834

0.838

0.819

0.824

TOP24 variables

0.820

0.822

0.824

0.819

0.824

0.837

0.840

0.818

0.824

0.820

0.821

0.834

0.837

0.817

0.821

TOP27 variables

0.827

0.817

0.818

0.827

0.829

0.839

0.841

0.822

0.827

0.814

0.820

0.835

0.838

0.814

0.819

TOP30 variables

0.828

0.812

0.817

0.821

0.822

0.838

0.840

0.819

0.823

nac

nac

nac

nac

nac

nac

ALL m variables a

0.826

0.795

0.811

nab

nab

0.840

0.841

0.701

0.725

nab

nab

0.838

0.839

0.713

0.725

 

(m = 51)

  

(m = 193)

  

(m = 200)

  1. am is the number of variables with presence in the GA solutions or with abs(coef) > 0 in the LASSO solution path. bnot calculated due to singularity.
  2. cno model with exactly 30 mutations.
  3. In bold the highest R2 per row is indicated.