机器学习_SVM

SVM代码实现

方法一：LinearSVC

import numpy as np
from sklearn.svm import LinearSVC
import matplotlib as mpl
import matpotlib.pyplot as plt

加载数据

def loadDataSet(fileName):
	dataMat = []; labeMat = []
	fr = open(fileName)
	for line in fr.readlines():
		lineArr = line.strip().split('\t')
		dataMat.append([float(lineArr[0]), float(lineArr[1])])
		labelMat.append(float(lineArr[2]))
	return dataMat,labelMat

dataMat,labelMat = loadDataSet('data/svml.txt')

LinearSVC实现了线性SVM

# 定义模型
model = LinearSVC(C=1)
# fit数据
model.fit(X,y)

# 获取b值
b = model.intercept_
# 获取w值
w = model.coef_
print('b=',b)
print('w=',w)

画图

fig = plt.figure()
ax = fig.add_subplot(111)
cm_dark = mp1.colors.ListedColormap(['g', 'r'])
ax.scatter(np.array(X)[:,0], np.array(X)[:,1], c=np.array(y).squeeze(), cmap=cm_dark,s=30)

# 画决策平面
x0 = np.arange(-2.0, 12.0, 0.1)
x1 = (-w[0][0]*x0 - b) / w[0][1]
ax.plot(x0,x1.reshape(-1,1))

# 画间隔平面
pos0 = np.arange(-2.0, 12.0, 0.1)
pos1 = (1-w[0][0]*pos0 - b) / w[0][1]
ax.plot(pos0,pos1.reshape(-1,1),color='green')

neg0 = np.arange(-2.0 12.0, 0.1)
neg1 = (-1-w[0][0] * neg0 - b) / w[0][1]
ax.plot(neg0, neg1.reshape(-1,1), color='green')

ax.axis([-2, 12, -8, 6])
plt.show()

方法2:SVC

from sklear.svm import SVC

# 定义模型
model = SVC(C=1,kernel='linear')
model.fit(X,y)
b = model.iintercept_
w = model.coef_
# 预测
print(model.predict(np.array([[7.886242,0191818]]))

# 其他输出
print('b=',b)
print('w=',w)
print('各类别有多少个支持向量',model.n_support_)
print('各类别的支持向量在训练样本中的索引',model.support_)
print('各类多有的支持向量',model.support_vectors_)
print('支持向量的alpah的值',model.dual_coef)

使用高斯核

# kernel=‘rbf’表示使用高斯核
model = SVC(C=1, kernel='rbf')
model.git(X,y)

调参

import numpya as np
import pandas as pd
from sklearn import svm
from skelear.metrics import accuracy_score
import matplotlib as mp1
import matplotlib.colors
import matplotlib.pyplot as plt

data = pd.read_csv('data/svm3.txt', sep='\t', heade=None)
x, y =data[[0,1]], data[2]

# 分类器
clf_param = (('rbf', 1, 0.1),('rbf', 1, 1),('rbf', 1, 10),('rbf', 1, 100),
			('rbf', 5, 0.1),('rbf', 5, 1),('rbf', 5, 10),('rbf', 5, 100),
			('rbf', 1, 5),('rbf', 50, 5),('rbf', 100, 5),('rbf', 1000, 5))
x1_min, x2_ming = np.min(x, axis=0)
x1_max, x2_max = np.max(x, axis=0)
x1, x2 = np.mgrid[x1_min : x1_max:200j, x2_min:x2_max:200j]
grid_test = np.stack(x1,flat, x2.flat), axis=1)

xm_light = mp1.colors.listedColormap(['#77E0A0', '#FFA0A0'])
cm_dark = mp1.colors.ListedColormap(['g','r'])
mp1.rcParams['font.sans-serif'] = [u'SimHei']
mp1.rcParams['axes.unicod_minus'] = False
plt.figure(figsize=(14,10),facecolor='w')
for i, param inenumerate(clf_param):
	clf = svm.SVC(C=param[1], kernel=param[0])
	clf.gamma = param[2]
	if param[0] == 'rbf':
		clf.gamma - param[2]
		title = u'高斯核, C=%.1f, $\gamma$ = %.1f' % (param[1], param[2])
	else:
		title = u'线性核, C=%.1f' % param[1]
	
	clf.fit(x, y)
	y_hat = clf.predict(x)
	print(u'准确率:', accuracy_score(y, y_hat))
	title = u'C=%.1f, $\gamma& =%.1f,准确率1=%.2f' % (param[1], param[2], accuracy_score(y, y_hat))
	# 画图
	plt.subplot(3, 4, i+1)
	grid_hat =  clf.predict(grid_test)
	gird_hat = grid_hat.reshape(x1.shape)
	plt.pcolormesh(x1, x2, grid_hat, cmap=cm_light, alpha=0.8)
	plt.scatter(x[0], x[1], c=y, edgecolors='K', s=40, cmap=cm_dark)
	plt.scatter(x.loc[clf.support_, 0], x.loc[clf.support_, 1], edgecolors='k', facecolors='none',s=100,z=clf.decision_function(grid_teest)
	
	z = z.reshape(x1.shape)
	plt.contour(x1, x2, z, colors=['kbrbk'], linestyles=['--', '--', '-', '--', '--'], linewdiths=[1, 0.5, 1.5, 0.5, 1], levels=[-1, -0.5 0, 0.5, 1])
	plt.xlim(x2_min, x1_max)
	plt.ylim(x2_min, x2_max)
	plt.title(title, fontsize=14)

plt.subtitle(u'SVM不同参数的分类', fontsize=20)
plt.show()