PCA

  PCA算法。PCA的目的是找到能够分离出最大方差的方向，所以首先求原来所有数据三个维度上的协方差，然后求这个协方差的特征值，最大特征值为第一个方向，从此以此类推。(PCA algorithm. The purpose of PCA is to find able to isolate the direction of maximum variance, so first find all the data in three dimensions on the original covariance, and then find the eigenvalues ​ ​ of the covariance, the biggest feature is the first in one direction, from and so on.)