sklearn SVM中dual_coef_为与支持向量对应的a*y

 dual_coef_返回的内容为

SVM对偶问题的所求得的支持向量对应的 拉格朗日系数与标签的乘机，即alpha_i * y_i

不要问为什么，看如下代码。

将dual_coef_处以对应标签y后剩下alpha，

显示 alpha大于等于0，且小于等于惩罚系数C。

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

# X, y = datasets.load_breast_cancer(return_X_y=True)
X, y = datasets.make_blobs(n_features=2,n_samples=200,centers=2, cluster_std=[2.0, 2.0], random_state=100)
y[y==0]=-1


X_train, X_test, y_train, y_test = train_test_split(X,y,test_size=0.2, random_state=100)
plt.scatter(X_train[:,0], X_train[:,1],c=y_train)
# plt.show()
m,d = X_train.shape
print("m::",m)
print("d::",d)
gamma = 1/(d*X_train.var())
model = SVC(kernel="rbf",gamma=gamma,C=1)

model.fit(X=X_train,y=y_train)


support = model.support_vectors_
plt.scatter(support[:,0], support[:,1],c='r',marker='*')
plt.show()

print(model.n_support_)
print(model.support_)
y_support = y_train[model.support_]
print(y_support)

ya = model.dual_coef_[0]

for lab, a in zip(y_support, ya):
    print(a/lab)

print(ya)


y_pred = model.predict(X=X_test)
Acc = accuracy_score(y_test, y_pred)
print("精度::",Acc)