Hierarchical Canonical Correlation Analysis Reveals Phenotype, Genotype, and Geoclimate Associations in Plants
Learning hierarchical organization of the datasets by HCCA. (a) The values of FT10 prediction by combinations of each pair of datasets are plotted against the condition number of the cross-covariance matrix of the two datasets. The -axis shows the log of the condition number of a pair of datasets (), and the -axis is the of the prediction obtained by SVR. (b) Comparison of the different hierarchical organizations of the datasets used for integration by CCA. The red combination is selected by the analysis of condition numbers in HCCA.