线性回归&&code

 1 # -*- coding: utf-8 -*-
 2 
 3 import numpy as np
 4 import matplotlib.pyplot as plt
 5 from certifi import __main__
 6 
 7 def cost(x,y,theta=np.zeros((2,1))):
 8    m=len(y);
 9    J=1.0/(2*m)*sum((x.dot(theta).flatten()-y)**2);
10    return J;
11 
12 def gradientDesc(x,y,theat=np.zeros((2,1)),alpha=0.001,iterations=1500):
13     m=len(y)
14     J=[]
15     for i in xrange(iterations):
16         a=theat[0][0]-alpha*(1.0/m)*sum((x.dot(theat).flatten()-y)*x[:,0]);
17         b=theat[1][0]-alpha*(1.0/m)*sum((x.dot(theat).flatten()-y)*1);
18         theat[0][0],theat[1][0]=a,b
19         print theat[0][0], theat[1][0]
20         print cost(x, y, theat);
21         
22     return theat;
23 
24 if __name__=="__main__":
25     x=np.array([[9,1],[15,1],[25,1],[14,1],[10,1],[18,1]]);
26     y=np.array([39,56,93,61,50,75]);
27     ans=gradientDesc(x, y);
28     xx=[1,30]
29     yy=[ans[0][0]*1+ans[1][0],ans[0][0]*30+ans[1][0]]
30     plt.plot(xx,yy)
31     plt.scatter(x[:,0],y)
32     plt.show()
33     
34     print 'end'
35 
36 #显示数据
37 '''
38 plt.scatter(x,y);
39 plt.show();
40 '''

结果显示

原文地址:https://www.cnblogs.com/wuxiangli/p/5879514.html