Classical dimension reduction algorithm -- princip
no vote
Application background PCA algorithm is mainly used to reduce the dimension, the data from high dimension to low dimension, simplify the expression of data. The specific algorithm steps are as follows:1, the sample matrix of the center of the sample matrix (matrix X each line is a sample)2, seeking covariance matrix3, the characteristic value, characteristic vector4, according to the contribution rate, to determine the number of feature vectors to form a transformation matrix5, take the former J column vector constitute the transformation matrix6, the sample matrix is projected onto the transform matrix, and the reduced dimension matrix is obtained.