Segmented principal component transform–principal component analysis

摘要: Abstract A new approach to perform Principal Component Analysis (PCA) on very wide matrices is proposed in this work. The procedure based an extension of the Transform (PCT) concept—the PCT being applied non-superimposed segments data matrix. It shown that method uses less memory than classical global PCA since decomposition done much smaller matrices, which has important impact requirements. also Segmented PCT-PCA (SegPCT-PCA) yields same results as performed by a PCA. This will allow study sets (e.g. 2D-NMR), were difficult do using approach. implementation SegPCT-PCA straightforward. An advantage it not necessary read complete matrix into main memory, could be for parallel calculations and cross-validation purposes.