[Mathematical model] The principal component analysis Method Python implementation

def PCA (Datamat, topnfeat=9999999):            #数据矩阵, output before Topnfeat feat    meanvals = Mean (Datamat, axis=0)           # Calculate Average    meanremoved = datamat-meanvals     Covmat = CoV (meanremoved, rowvar=0)        #计算协方差矩阵    eigvals,eigvects = Linalg.eig (Mat (Covmat)) #特征值,    eigvalind = Argsort (eigvals)               #排序, to find the eigenvalues. In fact, the most inconsistent with other changes    Eigvalind = eigvalind[:-( topnfeat+1): -1]   #反转    redeigvects = eigvects[:,eigvalind]        #    Lowddatamat = meanremoved * Redeigvects    #映射    reconmat = (Lowddatamat * redeigvects.t) + meanvals    return Lowddatamat, Reconmat

The mathematical principle of principal component analysis we can simply look at it: find the most varied direction as a new feature

If you want to infer the meaning of this division from the results of the program, Redeigvects is very critical, and it gives a mapping relationship

[Mathematical model] The principal component analysis Method Python implementation