? ? 當(dāng)你有一個(gè)數(shù)據(jù)集，每一條數(shù)據(jù)都M種屬性，然后你想知道M種屬性對數(shù)據(jù)集的影響的時(shí)候。你需要用到協(xié)方差矩陣。

? ? 求協(xié)方差矩陣之前請一定要知道協(xié)方差矩陣是干嘛的，是表示屬性之間關(guān)系的矩陣，協(xié)方差矩陣的規(guī)模只與屬性數(shù)量有關(guān)，和數(shù)據(jù)總量無關(guān)。blog.sciencenet.cn/blog-455004-805926.html 這里講的很清楚。

python代碼如下：

class PCA:

def avg(self,data):

avgData = [0]*len(data[0])

for i in range(0,len(data)):

for t in range(0,len(data[i])):

avgData[t] += data[i][t]

for i in range(0,len(avgData)):

avgData[i] = float(avgData[i])/len(data)

return avgData

def getCovMatrix(self,data,avg):

covData = [[0 for i in range(len(data[0]))] for i in range(len(data[0]))]

for i in range(0,len(data[0])):

for t in range(0,len(data[0])):

covData[i][t] = self.getCov(data,i,t,avg)

return covData

def getCov(self,data,col1,col2,avg):

cov = 0;

for i in range(0,len(data)):

cov += (data[i][col1]-avg[col1])*(data[i][col2]-avg[col2])

#print cov/(len(data)-1)

return cov/(len(data)-1)

data = [[-1,-1,1],[-2,-1,4],[-3,-2,-2],[1,1,1],[2,1,2],[3,2,1],[1,2,4]]

example = PCA()

avgdata = example.avg(data)

print example.getCovMatrix(data,avgdata)