Figure 8

Relevance of copy number variables to features extracted using regularized dual CCA and sparse CCA. The relevance of the copy number variables to the extracted features. For regularized dual CCA, the relevances are measured by the cross-loadings with the extracted features (left panel). For sparse CCA, the relevances are measured by the weights of the variables in the extracted features (right panel). The sparse method includes only a subset of the variables, necessary for explaining the information encoded in each feature. The regularized dual CCA (left panel) provides a relevance ranking of all variables. Note that only a subset of the original variable set was used in this analysis.