Fig 1.
(A) Comparison of F1-scores with the simulation data, (B) improvement of efficiency on the Hi-LASSO Python package and the Spark version, (C) comparison of F1-scores with the semi-real simulation data, and (D) robustness of feature selection.
Table 1.
Experimental results for feature selection performance (F1-scores) with the simulation data.
Table 2.
Experimental results for feature selection performance (F1-scores and numbers of non-zero variables in parentheses) with the semi-simulated data based on four TCGA cancer datasets.
The average of the experiments are shown, where bold-face indicates the best performance.