## 导入鸢尾花数据集
iris = datasets.load_iris()
data = iris.data[:, :2]
target = iris.target

## 区分训练集和测试集，75%的训练集和25%的测试集
train_data, test_data = train_test_split(np.c_[data, target])
## 训练并预测，其中选取k=15
clf = neighbors.KNeighborsClassifier(15, 'distance')
clf.fit(train_data[:, :2], train_data[:, 2])
Z = clf.predict(test_data[:, :2])
print '准确率:' ,clf.score(test_data[:, :2], test_data[:, 2])

colormap = dict(zip(np.unique(target), sns.color_palette()[:3]))
plt.scatter(train_data[:, 0], train_data[:, 1], edgecolors=[colormap[x] for x in train_data[:, 2]],c='', s=80, label='all_data')
plt.scatter(test_data[:, 0], test_data[:, 1], marker='^', color=[colormap[x] for x in Z], s=20, label='test_data')
plt.legend()
plt.show()