机器学习: KNN--python

今天介绍机器学习中比较常见的一种分类算法，K-NN，NN 就是 Nearest Neighbors， 也就是最近邻的意思，这是一种有监督的分类算法，给定一个 test sample, 计算这个 test sample 与 training set

里每个 training sample 的距离，选择离 test sample 最近的 K 个，然后通过投票选择这 K 个样本中，属于哪类的最多，那么这个 test sample 就属于哪类。K-NN 比较简单直观，也很好理解，一般需要考虑的就是设置 K 的大小，以及如何计算样本之间的距离，比较常用的是欧式距离。下面给出一段简单的代码，说明这个算法的使用。

from sklearn import datasets
import numpy as np
import operator

def Knn_Classify (x, Train_data, labels, k):
    N_sample = Train_data.shape[0]
    diff_mat = np.tile(x, (N_sample, 1)) - Train_data
    Sq_diffmat = diff_mat **2
    Sq_dis = Sq_diffmat.sum(axis = 1)
    Dis = Sq_dis ** 0.5
    Index = Dis.argsort()
    C_count = {}
    for i in range (k):
        votelabel = labels[Index[i]]
        C_count[votelabel] = C_count.get(votelabel, 0) + 1

    Sort_K = sorted(C_count.iteritems(),
       key = operator.itemgetter(1), reverse=True)

    return Sort_K

iris = datasets.load_iris()
x_data = iris.data
y_label = iris.target
class_name = iris.target_names

n_sample = len(x_data)

np.random.seed(0)
index = np.random.permutation(n_sample)
x_data = x_data[index]
y_label = y_label[index]

ratio = 0.8

train_x = x_data[ : int(ratio * n_sample)]
train_y = y_label[ : int(ratio * n_sample)]
test_x = x_data[int(ratio * n_sample) :]
test_y = y_label[int(ratio * n_sample) : ]

n_test = len(test_x)

p_label = np.zeros((len(test_y)))

for i in range (n_test):
    in_x = test_x [i, :]
    target_label = test_y [i]
    predict_value = Knn_Classify(in_x, train_x, train_y, 5)
    p_label[i] = predict_value[0][0]
#    print "the predict label is: ", predict_value
#    print "the target_label is: ", target_label

t = (p_label == test_y)
acc = t.sum()*1.0/len(test_y)

print "the accuracy is: ", acc