《机器学习实战》PCA降维

注释：由于各方面原因，理论部分不做介绍，网上很多自行百度吧！

pca.py

 1 import numpy as np
 2 import matplotlib.pyplot as plt
 3 import math
 4 
 5 def  loadDataSet(filename, delin = '	'):
 6     fr = open(filename)
 7     #读取分割存入数组
 8     stringArr = [line.strip().split(delin) for line in fr.readlines()]
 9     dataArr   = [list(map(float,line)) for line in stringArr]
10     return np.mat(dataArr)
11 def pca(dataMat, topNfeet = 9999999):
12     meanVals = np.mean(dataMat,axis=0)#求取平均值
13     meanRemoved = dataMat - meanVals
14     covMat = np.cov(meanRemoved,rowvar=0)#方差
15     eigVals, eigVects= np.linalg.eig(np.mat(covMat))#求解特征向量和特征值
16     eigValInd = np.argsort(eigVals)#对特征值进行排序
17     eigValInd = eigValInd[:-(topNfeet+1):-1]#最后的-1是防止越界的，当然你可以在前面加一个判断
18     redEigVects = eigVects[:,eigValInd]
19     lowDDataMat = meanRemoved*redEigVects #
20     reconMat = (lowDDataMat * redEigVects.T) + meanVals
21     return lowDDataMat, reconMat

main.py

 1 import PCA
 2 import matplotlib.pyplot as plt
 3 
 4 if __name__ == "__main__":
 5 
 6     dataMat = PCA.loadDataSet('testSet.txt')
 7     lowDMat, reconMat = PCA.pca(dataMat,1)
 8     fig = plt.figure()
 9     ax  = fig.add_subplot(111)
10     ax.scatter(dataMat[:,0].flatten().A[0],dataMat[:,1].flatten().A[0],marker = '^',s=90)
11     ax.scatter(reconMat[:,0].flatten().A[0],reconMat[:,1].flatten().A[0],marker = "o",s=50,c='red')
12     plt.show()

对丢失的值进行替代：

1 #零的数据都转化为平均值
2 def replaceNanWithMean():
3     dataMat = loadDataSet('secom.data',' ')
4     numFeat = dataMat.shape[1]
5     for i in range(numFeat):
6         meanVal = np.mean(dataMat[np.nonzero(~np.isnan(dataMat[:,i].A))[0],i])
7         dataMat[np.nonzero(np.isnan(dataMat[:,i].A))[0],i] = meanVal
8     return dataMat