机器学习测试笔记（13）——支持向量机（下）

def gamma_for_RBF():
       wine = datasets.load_wine()
#选取数据集前两个特征
       X = wine.data[:,:2]
       y = wine.target
       C = 1.0 #SVM参数正则化
       models  =(svm.SVC(kernel='rbf',gamma=0.1,C=C),
               svm.SVC(kernel='rbf',gamma=1,C=C),
               svm.SVC(kernel='rbf',gamma=10,C=C))
       models = (clf.fit(X,y)  for clf in models)
#设置标题
       titles = ('gamma =  0.1',
'gamma = 1',
'gamma = 10')
       fig, sub =  plt.subplots(1,3,figsize =(10,3))
       X0, X1 = X[:,0],X[:,1]
       xx, yy =  make_meshgrid(X0, X1)
#使用前面定义的函数画图
for clf, title, ax in  zip(models,titles,sub.flatten()):
              plot_contour(ax,clf,xx,yy,cmap=plt.cm.plasma,alpha=0.8)
              ax.scatter(X0,X1,c=y,cmap=plt.cm.plasma,s=20,edgecolors='k')
              ax.set_xlim(xx.min(),xx.max())  
              ax.set_ylim(yy.min(),yy.max())
              ax.set_xlabel('Feature  0')
              ax.set_ylabel('Feature  1')
              ax.set_xticks(())
              ax.set_yticks(())
              ax.set_title(title)
       plt.show()

from sklearn.pipeline import Pipeline
from sklearn.preprocessing import StandardScaler
from sklearn.model_selection import train_test_split
def SVM_for_boston():
   boston =  datasets.load_boston()
   X,y =  boston.data,boston.target
    X_train,X_test,y_train,y_test = train_test_split(X, y, random_state  =8)
for kernel in  ['linear','rbf','sigmoid','poly']:
      svr =  svm.SVR(kernel=kernel)
       svr.fit(X_train,y_train)
      print(kernel,'核函数的模型训练集得分:  {:.3f}'.format(svr.score(X_train,y_train)))
      print(kernel,'核函数的模型测试集得分: {:.3f}'.format(svr.score(X_test,y_test)))

linear 核函数的模型训练集得分: 0.709
linear 核函数的模型测试集得分: 0.696
rbf 核函数的模型训练集得分:  0.192
rbf 核函数的模型测试集得分:  0.222
sigmoid 核函数的模型训练集得分: 0.059
sigmoid 核函数的模型测试集得分: 0.075
poly 核函数的模型训练集得分: 0.195
poly 核函数的模型测试集得分: 0.207

ax = plt.gca()
#将特征数字中的最大和最小值一散点图形式画出来
       plt.plot(X.min(axis=0),'v',label='min')
       plt.plot(X.max(axis=0),'^',label='max')
       plt.yscale('log') #纵坐标为对数形式
       plt.legend(loc='best') #设置图注位置最佳
       ax.set_xlabel('feature')
       ax.set_ylabel('featuremagnitude')
       plt.show()

#对训练集和测试集数据进行预处理
scaler = StandardScaler()
       scaler.fit(X_train)
       X_train_scaler =scaler.transform(X_train)
       X_test_scaler = scaler.transform(X_test)
       plt.plot(X_train_scaler.min(axis=0),'v',label='trainset min')
       plt.plot(X_train_scaler.max(axis=0),'^',label='trainset max')
       plt.plot(X_test_scaler.min(axis=0),'v',label='testset min')
       plt.plot(X_test_scaler.max(axis=0),'^',label='testset max')
       plt.legend(loc='best')
       ax.set_xlabel('scaledfeature')
       ax.set_ylabel('scaledfeature magnitude')
       plt.show()

linear 预处理后训练集得分: 0.706
linear 预处理后测试集得分: 0.698
rbf 预处理后训练集得分:  0.665
rbf 预处理后测试集得分:  0.695
sigmoid 预处理后训练集得分: 0.564
sigmoid 预处理后测试集得分: 0.634
poly 预处理后训练集得分: 0.686
poly 预处理后测试集得分: 0.623

# 调节参数
for kernel in ['linear','rbf','sigmoid','poly']:
svr = svm.SVR(kernel=kernel,C=100,gamma=0.1)
svr.fit(X_train_scaler,y_train)
print(kernel,'调节参数后训练集得分: {:.3f}'.format(svr.score(X_train_scaler,y_train)))
print(kernel,'调节参数后测试集得分:{:.3f}'.format(svr.score(X_test_scaler,y_test)))

linear 调节参数后训练集得分: 0.706
linear 调节参数后测试集得分: 0.699
rbf 调节参数后训练集得分:  0.966
rbf 调节参数后测试集得分:  0.894
sigmoid 调节参数后训练集得分: -3768.620
sigmoid 调节参数后测试集得分: -5454.605
poly 调节参数后训练集得分: 0.950
poly 调节参数后测试集得分: 0.318

# 调节参数
for kernel in ['linear','rbf','sigmoid','poly’]:
  svr =svm.SVR(kernel=kernel,C=100,gamma=0.005)
 svr.fit(X_train_scaler,y_train)
  print(kernel,'调节参数（C=100,gamma=0.005）后训练集得分:{:.3f}'.format(svr.score(X_train_scaler,y_train)))
  print(kernel,'调节参数（C=100,gamma=0.005）后测试集得分: {:.3f}'.format(svr.score(X_test_scaler,y_test)))

linear 调节参数后训练集得分: 0.706
linear 调节参数后测试集得分: 0.699
rbf 调节参数后训练集得分:  0.841
rbf 调节参数后测试集得分:  0.829
sigmoid 调节参数后训练集得分: 0.698
sigmoid 调节参数后测试集得分: 0.695
poly 调节参数后训练集得分: 0.240
poly 调节参数后测试集得分: 0.232