Skip to main content
Fig. 2 | BMC Bioinformatics

Fig. 2

From: Data integration by multi-tuning parameter elastic net regression

Fig. 2

Factors associated with the change of optimal penalty ratio parameter κ. a For the scenario with different numbers of informative features, q1 < q2, κ increases monotonically to 1 (i.e. less differential penalization) as the effect size in the first omic type increases. b For fixed effect sizes β1 and β2, κ becomes smaller (more differential penalization) as the number of informative features in the second omic type increases. c As the overall proportion of informative features of both types, \( \frac{q_1+{q}_2}{p} \), decreases relative to the total number of features, κ approaches 1, i.e. less differential penalization is required to maximize the AUC. Dots represent the optimal weights and caps represent the standard error of the mean; N = 200 simulated data sets

Back to article page