From: A structured overview of simultaneous component based data integration
Principle | Methods aiming at this principle |
---|---|
Same weight for all matrices (naive approach) | SCA-P |
More weight for smaller matrices | SUM-PCA, MFA |
More weight for less redundant matrices | MFA |
More weight for matrices with more stable predictive information | PCovR |
More weight for matrices that share more information with other matrices (K > 2) | STATIS |