主成分分析法的数值实现算法

The numerical algorithm of principal component analysis

本文通过Householder变换将已知信息的相关矩阵B化成三对角矩阵,给出了确定其特征值及相应的特征向量的方法,从而得到了用主成分分析法分析问题的一种数值算法。

A method to find eigenvalues and corresponding eigenvectors of the known information corre- sponding matrix was given through changing it into tfidiagonal matrix by Householder. A kind of numerical computation method of the principal component analysis was obtained.