update

sherlock-coding · sherlock-coding · commit 8f67fa7d5eb4 · 2014-12-16T19:53:27.000+08:00
diff --git a/Homework/HousingPricePrediction.py b/Homework/HousingPricePrediction.py
@@ -2,6 +2,27 @@
 import math
 import matplotlib.pyplot as plt
 
+#特征缩放
+def FeatureScaling(X,n):
+    X_fs = X
+    avg_fs = []
+    s_fs = []
+    for i in range(0,n-1):
+        x_i = []
+        for x in X:
+            x_i.append(x[i])
+        avg = sum(x_i)/(len(X))
+        s = max(x_i)- min(x_i)
+        avg_fs.append(avg)
+        s_fs.append(s)
+        for j in range(0,len(X)):
+            x_i[j] = (x_i[j]-avg)/s
+        for k in range(0,len(X)):
+            X[k][i] = x_i[k]
+            
+    return X_fs,avg_fs,s_fs
+    
+
 #求h(x,theta)
 def GetHThetaX(x,theta):
     theta_1 = theta[1:]
@@ -29,17 +50,23 @@ def GetGradient(X,Y,theta,j):
 
 
 #n表示n维的特征向量
-def LMS_Prediction(X,X1,Y,n,year):
+def LMS_theta(X,Y,n):
         #theta全都初始化为0
         #theta = [i for i in range(0,n)]
-        theta = [-1500,1]
+        theta = [0]*n
+
+        X1 = []
+        for x in X:
+            X1.append(x[0])
+
+        print(X1)
         
         plt.plot(X1,Y,'ro')
-        plt.axis([2000,2015,1,15])
+        plt.axis([-0.8,0.8,1,15])
         plt.plot(X1,[theta[0]+theta[1]*x for x in X1],'y-')
 
         #learning rate设置
-        alpha = 0.00000001
+        alpha = 0.1
 
         steps = 0
 
@@ -66,13 +93,20 @@ def LMS_Prediction(X,X1,Y,n,year):
         plt.plot(X1,[theta[0]+theta[1]*x for x in X1],'r-')
         plt.show()
         
+        return theta
+        
         
 
 if __name__ == '__main__':
     X = [[2000],[2001],[2002],[2003],[2004],[2005],[2006],[2007],[2008],[2009],[2010],[2011],[2012],[2013]]
     Y = [2.000,2.500,2.900,3.147,4.515,4.903,5.365,5.704,6.853,7.971,8.561,10.000,11.280,12.900]
-    X1 = [2000,2001,2002,2003,2004,2005,2006,2007,2008,2009,2010,2011,2012,2013]
-    LMS_Prediction(X,X1,Y,2,2014)
+    X_fs,avg_fs,s_fs = FeatureScaling(X,2)
+    theta = LMS_theta(X_fs,Y,2)
+    print(theta)
+    x_p = [2014]
+    for i in range(0,2-1):
+        x_p[i] = (x_p[i]-avg_fs[i])/s_fs[i]
+    print(GetHThetaX(x_p,theta))
     #X = [[1],[2],[5],[7], [10], [15]]
     #Y = [2, 6, 7, 9, 14, 19]
     #X = [[2000],[2001],[2002],[2003],[2004],[2005],[2006],[2007],[2008],[2009],[2010],[2011],[2012],[2013]]
