test_non_linear.py

# coding: utf8
# svm/test_non_linear.py
import smo
import numpy as np
from sklearn import datasets
from scipy.io import loadmat
import matplotlib.pyplot as plt

data = loadmat('data/ex6data2.mat')

X = np.mat(data['X'])
y = np.mat(data['y'], dtype=np.float)
y[y==0] = -1

m, n = X.shape
C = 1.0
tol = 1e-3
maxIter = 5
kernel = smo.rbfKernel(0.1)

trainSimple, train, predict = smo.getSmo(X, y, C, tol, maxIter, kernel=kernel)
alphas, w, b, supportVectorsIndex, supportVectors, iterCount = train()
print supportVectorsIndex
print len(supportVectorsIndex)
print 'iterCount:%d' % iterCount

predictions = predict(X, alphas, b, supportVectorsIndex, supportVectors)
errorCount = (np.multiply(predictions, y).A  < 0 ).sum()
print errorCount
print 'error rate: %.2f'%(float(errorCount)/m)

# 绘制数据点
x1Min = X[:, 0].min()
x1Max = X[:, 0].max()
x2Min = X[:, 1].min()
x2Max = X[:, 1].max()
plt.title('C=%.1f'%C)
plt.xlabel('X1')
plt.ylabel('X2')
plt.xlim(x1Min, x1Max)
plt.ylim(x2Min, x2Max)

for i in range(m):
    x = X[i].A[0]
    if y[i] == 1:
        color = 'black'
        if i in supportVectorsIndex:
            color = 'red'
        plt.scatter(x[0], x[1], marker='*', color=color, s=50)
    else:
        color = 'green'
        if i in supportVectorsIndex:
            color = 'red'
        plt.scatter(x[0], x[1], marker='o', color=color, s=50)


# 绘制决策边界
xx1, xx2 = np.meshgrid(
    np.linspace(x1Min, x1Max, 100),
    np.linspace(x2Min, x2Max, 100)
)
predictX = np.mat(np.c_[xx1.ravel(), xx2.ravel()])
predictions = predict(predictX, alphas, b, supportVectorsIndex, supportVectors)
predictions = predictions.reshape(xx1.shape)
plt.contour(xx1, xx2, predictions, [0.5], linewidths=5)
plt.show()