knn分类算法底层实现（python)

# coding:utf-8
from collections import defaultdict

import numpy as np
from numpy import *


class knn:
    def __init__(self):
        pass

    ##给出训练数据以及对应的类别
    def createDataSet(self):
        group = array([[1.0, 2.0], [1.2, 0.1], [0.1, 1.4], [0.3, 3.5]])
        labels = ['A', 'A', 'B', 'B']
        return group, labels

    ###通过KNN进行分类
    def classify(self, input, data_set, labels, k):
        # 将input扩展成 n行的举证
        in_matrix = tile(input, (len(data_set), 1))
        # 输入举证和数据集做差 (x1-x2)
        diff = in_matrix - data_set
        # (x1-x2)**2
        diff = diff ** 2
        # (x1-x2）** 2 +（y1-y2)**2
        sm = np.sum(diff, axis=1)
        sm = np.sqrt(sm)
        # 将距离排序
        si = np.argsort(sm)
        label_dict = defaultdict(int)
        max_num = 0
        target_lb = None
        for i in range(k):
            index = si[i]
            lb = labels[index]
            label_dict[lb] += 1
            if label_dict[lb] > max_num:
                max_num = label_dict[lb]
                target_lb = lb
        return target_lb

    def norm_data_set(self, data_set):
        '''
        将数据集归一化
        :param data_set: 
        :return: 
        '''
        # 最大和最小的行向量
        val_min = np.min(data_set, 0)
        val_max = np.max(data_set, 0)
        # 数据变动范围向量
        val_range = val_max - val_min
        row_num = data_set.shape[0]
        matrix_range = tile(val_range, (row_num, 1))
        matrix_sp = data_set - tile(val_min, (row_num, 1))
        matrix_normal = matrix_sp / matrix_range
        return matrix_normal

    def norm_vec(self, vec, data_set):
        '''
        将被分类的向量归一化
        :param vec: 
        :param data_set: 
        :return: 
        '''
        data_set = np.vstack((data_set, vec))
        # 最大和最小的行向量
        val_min = np.min(data_set, 0)
        val_max = np.max(data_set, 0)
        # 数据变动范围向量
        val_range = val_max - val_min
        span = vec - val_min
        return span / val_range


if __name__ == '__main__':
    k = knn()
    g, l = k.createDataSet()
    ng = k.norm_data_set(g)
    vec = [0.3, 3.2]
    n_vec = k.norm_vec(vec, g)
    b = k.classify(n_vec, ng, l, 4)