Skip to main content

Table 2 Confusion matrix when linear regression, lasso and rf were applied to the synthetic dataset (\(N=1000,N_1=10,M=10,K=3\))

From: Novel feature selection method via kernel tensor decomposition for improved multi-omics data analysis

  Linear regression Lasso Rf
Adjusted Adjusted     
\(P_i \le 0.01\) \(P_i > 0.01\) Selected not selected Selected Not selected
\(k=1\)       
\(i \le 2N_1\) 0.07 19.93 4.62 15.383 17.55 (5.82) 2.45 (14.18)
\(i > 2N_1\) 0.03 979.97 2.12 977.88 495.43 (14.18) 484.57 (965.82)
\(k=2\)       
\(i \le N_1\), \(2N_1< i \le 3N_1\) 0.07 19.93 4.70 15.30 17.69 (5.67) 2.31 (14.33)
Other than above 0.01 979.99 2.27 977.73 494.70 (14.33) 485.30 (965.67)
\(k=3\)       
\(i \le N_1,3N_1< i \le 4N_1\) 0.09 19.91 4.55 15.45 17.71(5.46) 2.29 (14.54)
Other than above 0.01 979.99 2.12 977.78 496.68 (14.54) 483.32 (965.46)
  1. For cases when rf was employed, the results when the top most \(2 N_1\) features with larger absolute importance were selected have also been shown in parentheses