学习笔记：sklearn-SVM

查看svm.SCV的不同kernel在moons、circles、blob、classification数据上的表现
kernel= linear/poly 为线性核
kernel= rbf/sigmoid 为非线性核

import numpy as np
import matplotlib.pyplot as plt
from sklearn import svm
from sklearn.datasets import make_circles, make_moons, make_blobs, make_classification

n_sample=100

datasets=[
    make_moons(n_sample),
    make_circles(n_sample, noise=0.2, factor=0.5),
    make_blobs(n_sample, centers=2),
    make_classification(n_samples=n_sample, n_features=2, n_informative=2, n_redundant=0)
]

kernel=['linear', 'poly', 'rbf', 'sigmoid']

for idx ,(X, y) in enumerate(datasets):
    plt.subplot(idx+1,5,1)
    plt.scatter(X[:,0],X[:,1], c=y)
    for i in range(len(kernel)):
        clf=svm.SVC(kernel=kernel[i], gamma='auto').fit(X, y)
        score=clf.score(X,y) 
        print(kernel[i],score)
        y_pred=clf.predict(X)
        plt.subplot(idx+1,5,i+2)
        plt.scatter(X[:,0],X[:,1], c=y_pred)
    plt.show()

调整松弛系数C

data=make_moons(n_sample)
X, y=data[0], data[1]
c_range=[1,5,10,20]
for i in c_range:
    clf_soft=svm.SVC(kernel='rbf', gamma='auto', C=i).fit(X, y)
    score=clf_soft.score(X,y)
    print(score)

样本数量不均衡时，设置类别权重class_weight

import numpy as np
import matplotlib.pyplot as plt
from sklearn import svm
from sklearn.datasets import make_blobs

pos_sample, neg_sample=200, 100

X, y=make_blobs(n_samples=[pos_sample, neg_sample],
               centers=[[0.,0.],[2.,2.]])
plt.scatter(X[:, 0], X[:,1],c=y)
# plt.show()

clf=svm.SVC(kernel='rbf', C=1.0, class_weight='balanced').fit(X, y)
#svm.SVC(kernel='linear', class_weight={1:2})
clf.score(X,y)