Skip to main content
Figure 3 | BMC Bioinformatics

Figure 3

From: A flexible framework for sparse simultaneous component based data integration

Figure 3

Boxplots of the proportion of recovered variables: 50 percent of true zeros. Boxplots of the proportion of variables correctly classified (selected versus dropped) in function of the value of the tuning parameters. Case with 50 percent of the variables dropped. The different panels refer to the different combinations of structure in the variable selection (from top to bottom: within blocks, between blocks, within and between blocks) and of sparseness approach (from left to right: lasso, group lasso, Elitist lasso, and sparse group lasso). The panels referring to the sparse group lasso are with varying values for the lasso tuning parameter and with the group lasso tuning parameter fixed at f = 10.

Back to article page