-
Notifications
You must be signed in to change notification settings - Fork 1
Expand file tree
/
Copy pathpca.py
More file actions
71 lines (48 loc) · 1.99 KB
/
Copy pathpca.py
File metadata and controls
71 lines (48 loc) · 1.99 KB
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
# -*- coding: utf-8 -*-
"""
Created on Thu Dec 12 19:42:23 2016
@author:june
code of PCA Algrithom
"""
import numpy as np
from util.tools import scale
class PCA():
def __init__(self, percent=0.99, rowvar=0):
self.percent = percent # 根据要求的方差百分比，求出所需要的特征值的个数n
self.rowvar = rowvar #若rowvar非0，一列代表一个样本，为0，一行代表一个样本
def percent2n(self, eigVals):
sortArray = np.sort(eigVals) # 升序
sortArray = sortArray[-1::-1] # 逆转，即降序
arraySum_needed = sum(sortArray) * self.percent
tmp = 0
num = 0
for i in sortArray:
tmp += i
num += 1
if tmp >= arraySum_needed:
return num
def pca(self, dataMat):
newData= scale(dataMat)
covMat = np.cov(newData, rowvar=self.rowvar) # 求协方差矩阵,return ndarray；
eigVals, eigVects = np.linalg.eig(np.mat(covMat)) # 求特征值和特征向量,特征向量是按列放的，即一列代表一个特征向量
n = self.percent2n(eigVals) # 要达到percent的方差百分比，需要前n个特征向量
eigValIndice = np.argsort(eigVals) # 对特征值从小到大排序
n_eigValIndice = eigValIndice[-1:-(n + 1):-1] # 最大的n个特征值的下标
n_eigVect = eigVects[:, n_eigValIndice] # 最大的n个特征值对应的特征向量
lowDDataMat = newData * n_eigVect # 低维特征空间的数据
reconMat = (lowDDataMat * n_eigVect.T) + np.mean(dataMat) # 重构数据
return lowDDataMat, reconMat
def demo():
data = np.loadtxt("testSet.txt")
pca = PCA(rowvar=0)
lowdata, reconmat = pca.pca(data)
print(lowdata.size)
def sklearn_demo():
from sklearn.decomposition import PCA
data = np.loadtxt("testSet.txt")
pca = PCA(2)
pca.fit(data)
print(pca.explained_variance_ratio_)
if __name__ == "__main__":
#demo()
sklearn_demo()