24/8/6算法笔记 不同核函数

import numpy as np
from sklearn import datasets
from sklearn.svm import SVC
from sklearn.model_selection import train_test_split
from sklearn.metrics import accuracy_score
import matplotlib.pyplot as plt

加载数据

X,y=datasets.load_wine(return_X_y = True)
display(X.shape)

X_train,X_test,y_train,y_test = train_test_split(X,y,test_size=0.2)
X_train.shape

建模

线性linear

svc = SVC(kernel = 'linear')
          
svc.fit(X_train,y_train)
          
y_pred = svc.predict(X_test)
    
score = accuracy_score(y_test,y_pred)
print('使用核函数为linear，得分为：',score)

#二维：shape (3,13),
#三分类问题-->三个方程
#特征13个，所以系数是13
svc.coef_
svc.intercept_

poly多项式(方程幂次大于1）

svc = SVC(kernel = 'poly')#升维，数据由少变多
  
svc.fit(X_train,y_train)
          
y_pred = svc.predict(X_test)
    
score = accuracy_score(y_test,y_pred)

print('使用核函数为linear，得分为：',score)

rbf高斯分布,正态分布

svc = SVC(kernel = 'rbf')#默认的，一般这种核函数效果好，属于正态分布
  
svc.fit(X_train,y_train)
          
y_pred = svc.predict(X_test)
    
score = accuracy_score(y_test,y_pred)

print('使用核函数为linear，得分为：',score)

sigmoid函数

svc = SVC(kernel = 'sigmoid')
  
svc.fit(X_train,y_train)
          
y_pred = svc.predict(X_test)
    
score = accuracy_score(y_test,y_pred)

print('使用核函数为linear，得分为：',score)

非线性核函数

from matplotlib.colors import ListedColormap

创造数据

X,y= datasets.make_circles(n_samples=100,factor=0.7)

X +=np.random.randn(100,2)*0.03
display(X.shape,y.shape)

plt.figure(figsize= (5,5))
cmap = ListedColormap(colors= ['blue','red'])
plt.scatter(X[:,0],X[:,1],c=y,cmap = cmap)

线性核函数

svc = SVC(kernel = 'linear')
  
svc.fit(X,y)
          
svc.score(X,y)

多项式poly(升维)

svc = SVC(kernel = 'poly',degree=2)#二次幂
  
svc.fit(X,y)
          
svc.score(X,y)

高斯核函数rbf

svc = SVC(kernel = 'rbf')
  
svc.fit(X,y)
          
svc.score(X,y)